Ecological Monographs, 75(1), 2005, pp. 37–63
q 2005 by the Ecological Society of America
ECOSYSTEM EFFECTS OF BIODIVERSITY MANIPULATIONS IN
EUROPEAN GRASSLANDS
E. M. SPEHN,1,24 A. HECTOR,2,14 J. JOSHI,3 M. SCHERER-LORENZEN,4,15 B. SCHMID,3 E. BAZELEY-WHITE,2
C. BEIERKUHNLEIN,5 M. C. CALDEIRA,6 M. DIEMER,3 P. G. DIMITRAKOPOULOS,7 J. A. FINN,8,16 H. FREITAS,6,17
P. S. GILLER,8 J. GOOD,8 R. HARRIS,8 P. HÖGBERG,9 K. HUSS-DANELL,10 A. JUMPPONEN,9,10,18 J. KORICHEVA,11
P. W. LEADLEY,1,19 M. LOREAU,12 A. MINNS,2 C. P. H. MULDER,9,10,20 G. O’DONOVAN,8,21 S. J. OTWAY,2
C. PALMBORG,10 J. S. PEREIRA,6 A. B. PFISTERER,3 A. PRINZ,5,22 D. J. READ,13 E.-D. SCHULZE,4
A.-S. D. SIAMANTZIOURAS,7 A. C. TERRY,13 A. Y. TROUMBIS,7 F. I. WOODWARD,13 S. YACHI,12,23
AND J. H. LAWTON 2
1Institute of Botany, University of Basel, Schoenbeinstrasse 6, Basel, Switzerland, CH-4056
Natural Environmental Research Council (NERC) Centre for Population Biology, Imperial College London,
Silwood Park Campus, Ascot, Berkshire, UK, GB-SL5 7PY
3Institute of Environmental Sciences, University of Zurich, Winterthurerstrasse 190, Zurich, Switzerland, CH-8057
4Max-Planck-Institute for Biogeochemistry, Postfach 10 01 64, Jena, Germany, D-07701
5Lehrstuhl Biogeographie, Universität Bayreuth, Bayreuth, Germany, D-95440
6Departmentos de Engenharia Florestal, Universidade Tecnica de Lisboa, Tapada da Ajuda, Lisboa, Portugal, PT-1399
7Biodiversity Conservation Laboratory, Department of Environmental Studies, University of the Aegean, Mytilene, Lesbos,
Greece, GR-811 00
8Department of Zoology, Ecology and Plant Science, University College Cork, Lee Maltings, Prospect Row, Cork, Ireland
9Department of Forest Ecology, Swedish University of Agricultural Sciences, Umeå, Sweden, SE-90183
10Crop Science Section, Department of Agricultural Research for Northern Sweden, Box 4097,
Swedish University of Agricultural Sciences, Umeå, Sweden, SE-90403
11Section of Ecology, Department of Biology, University of Turku, 20014 Turku, Finland
12Laboratoire d’Écologie, UMR 7625, École Normale Supérieure, 46 Rue d’Ulm, Paris Cedex 05, France, FR-75230
13Department of Animal and Plant Sciences, University of Sheffield, South Yorkshire, UK, GB-S10 2TN
2
Abstract. We present a multisite analysis of the relationship between plant diversity
and ecosystem functioning within the European BIODEPTH network of plant-diversity
manipulation experiments. We report results of the analysis of 11 variables addressing
several aspects of key ecosystem processes like biomass production, resource use (space,
light, and nitrogen), and decomposition, measured across three years in plots of varying
plant species richness at eight different European grassland field sites. Differences among
sites explained substantial and significant amounts of the variation of most of the ecosystem
processes examined. However, against this background of geographic variation, all the
aspects of plant diversity and composition we examined (i.e., both numbers and types of
species and functional groups) produced significant, mostly positive impacts on ecosystem
processes.
Analyses using the additive partitioning method revealed that complementarity effects
(greater net yields than predicted from monocultures due to resource partitioning, positive
interactions, etc.) were stronger and more consistent than selection effects (the covariance
between monoculture yield and change in yield in mixtures) caused by dominance of species
with particular traits. In general, communities with a higher diversity of species and functional groups were more productive and utilized resources more completely by intercepting
Manuscript received 20 October 2003; revised 29 June 2004; accepted 12 July 2004. Corresponding Editor: N. C. Kenkel.
Present address: Institute of Environmental Sciences, University of Zürich, Winterthurerstrasse 190, Zürich, Switzerland,
CH-8057.
15 Present address: Swiss Federal Institute of Technology Zürich (ETH), Institute of Plant Sciences, Universitätsstrasse 2,
Zürich, Switzerland, CH-8092.
16 Present address: Teagasc Environmental Research Centre, Johnstown Castle, Wexford, Ireland.
17 Present address: Departamento de Botânica, Universidade de Coimbra, 3000 Coimbra, Portugal.
18 Present address: Division of Biology, Kansas State University, Manhattan, Kansas 66506 USA.
19 Present address: Ecologie des Populations et Communautés, Université Paris Sud XI, URA CNRS 2154, Bâtiment 326,
Orsay Cedex, France, FR-91405.
20 Present address: Department of Biology and Wildlife and Institute of Arctic Biology, University of Alaska, 410 Irving
I, Fairbanks, Alaska 99775-7000 USA.
21 Present address: Department of Environmental Resource Management, University College of Dublin, Belfield, Dublin,
Ireland.
22 Present address: Landesbund für Vogelschutz, Hilpoltstein, Germany, D-91157.
23 Present address: Research Institute for Humanity & Nature (RINH), Kamigyo-ku, Kyoto 606-0878, Japan.
24 E-mail: eva.spehn@unibas.ch
14
37
E. M. SPEHN ET AL.
38
Ecological Monographs
Vol. 75, No. 1
more light, taking up more nitrogen, and occupying more of the available space. Diversity
had significant effects through both increased vegetation cover and greater nitrogen retention
by plants when this resource was more abundant through N2 fixation by legumes. However,
additional positive diversity effects remained even after controlling for differences in vegetation cover and for the presence of legumes in communities. Diversity effects were
stronger on above- than belowground processes. In particular, clear diversity effects on
decomposition were only observed at one of the eight sites.
The ecosystem effects of plant diversity also varied between sites and years. In general,
diversity effects were lowest in the first year and stronger later in the experiment, indicating
that they were not transitional due to community establishment. These analyses of our
complete ecosystem process data set largely reinforce our previous results, and those from
comparable biodiversity experiments, and extend the generality of diversity–ecosystem
functioning relationships to multiple sites, years, and processes.
Key words: BIODEPTH, European plant-experiment network; biomass production; canopy structure; complementarity effect; decomposition; legumes; light use; multisite grasslands field study; nitrogen retention; plant species richness; resource use; selection effect.
INTRODUCTION
The last decade has seen the intensified collaboration
of population and community ecologists with physiological and ecosystem ecologists in the study of the
relationship between biodiversity and ecosystem functioning (Schulze and Mooney 1993, Loreau et al. 2001,
2002, Kinzig et al. 2002). One of the main motivations
for this research is the change in and loss of diversity
seen as a result of human activities and climate change.
Identifying general patterns in a research area such as
this can be a major obstacle because even well-replicated ecological studies are often conducted at single
points in space and time, and often focus on one or a
small number of variables. This paper reports the results of a large-scale pan-European project (BIODEPTH—BIODiversity and Ecological Processes in
Terrestrial Herbaceous ecosystems) that examined the
relationship between plant diversity (species richness
and functional-group richness of vascular plants) and
ecosystem functioning in experimental grassland communities based on locally and naturally occurring species, established from seed under field conditions. A
major aim of BIODEPTH was to explicitly address
whether biodiversity effects on ecosystem functioning
are consistent in space and time, over a range of climate
and soil conditions, and with different local species
pools. It is well known that natural variation has a
major influence on ecosystem functioning (Schulze et
al. 1995, Grime 1998, Huston and McBride 2002). In
this experiment, we addressed the extent to which diversity affects ecosystem functioning at individual sites
and the extent to which diversity interacts with variation in ecosystem functioning across sites (Hector et
al. 1999, Fridley 2002).
Biodiversity effects on ecosystem processes could
vary with environmental conditions if individual species performances and interactions are differentially
affected by environmental variation such as changes in
soil fertility and light availability (Tilman 1988, Bazzaz
1996, Fargione and Tilman 2002). However, if differential responses to environmental variation are aver-
aged out in multi-species communities this may increase the consistency of the biodiversity–ecosystemfunctioning relationships across environmental gradients (Cropp and Gabric 2002). Only a few multisite
biodiversity experiments have been carried out so far
(BIODEPTH; CLUE [Changing land usage, enhancement of biodiversity and ecosystem development], Van
der Putten et al. 2000; Emmerson et al. 2001), which
try to identify the role of extrinsic factors (Tilman
1994, Fridley 2003) that vary and act at larger scales
than the commonly investigated intrinsic factors (including community processes) that act at a local scale
within systems.
Our own multisite analyses so far have focused mainly on aboveground biomass production (Hector et al.
1999, 2002a) from the second year of the experiment.
We reported that the overall relationship between diversity and biomass production was positive and log
linear, but that slopes from individual field sites ranged
from positive to zero (Hector et al. 1999). Reviews of
comparable experiments with plants (Schmid et al.
2002a) report relationships that all fall within this range
from positive and linear (e.g., Tilman et al. 2001, He
et al. 2002), to approximately log linear (e.g., Tilman
et al. 1996, 1997a, Niklaus et al. 2001, Reich et al.
2001, van Ruijven and Berendse 2003) to slopes not
significantly different from zero (e.g., Hooper and Vitousek 1997).
One major mechanism that could allow production
to increase with diversity is complementary use of resources, such as through light partitioning, in which
differences in morphology between species enhance the
structural complexity of the vegetation and allow for
more complete utilization of PAR (photosynthetic active radiation; Hirose and Werger 1994, Naeem et al.
1994, 1995, Spehn et al. 2000b, Fridley 2003). Diversity is also likely to increase complementarity in nutrient uptake, either by different species acquiring nutrients from different portions of the available pool in
space, time, or chemical form, thereby increasing, e.g.,
total nitrogen retention (Spehn et al. 2002) and de-
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
creasing losses to leaching (Scherer-Lorenzen et al.
2003).
Biodiversity is not a one-dimensional ecological variable that acts in a simple way on ecosystem processes
but is instead multifaceted, with complex effects. This
makes the statistical analysis and interpretation of biodiversity experiments complex and even the proximate
mechanisms difficult to isolate. Several major issues
have already been identified that require special attention in analyses:
1) Effects of biodiversity in experimental communities can be separated into two major categories of
explanation: niche complementarity (e.g., resource partitioning), plus positive (facilitative) interactions (Hector 1998, Loreau and Hector 2001) vs. ‘‘sampling effects’’ resulting from the greater probability of including a species or combinations of species with particular
traits that dominate randomly assembled high-diversity
communities (Huston 1997, Loreau et al. 2001). We
employ the additive partitioning method of Loreau and
Hector (2001) because it allows us to quantitatively
define an overall net biodiversity effect and then separate this into a selection effect and a complementarity
effect (see Methods: Additive partitioning . . . , below),
whereas previous discussions have often treated the
two mechanisms as mutually exclusive alternatives
(Huston 1997, Huston et al. 2000).
2) One of the major functional traits of plants that
affects many ecosystem processes is the ability to symbiotically fix atmospheric nitrogen. Hence, one of the
major interactions likely to be of functional importance
is the one between N fixers and non-fixing species.
Whilst our experiment was not designed to isolate the
role of legumes in generating biodiversity effects, we
conducted analyses to test whether legumes formed a
significant component of compositional diversity and
whether additional species-richness effects existed beyond the effect of the presence of legume species.
3) Biodiversity effects on ecosystem functioning
may be synergistic and simultaneously affect several
interrelated ecosystem processes. For example, diverse
communities may establish faster, as it is more likely
that other species can compensate an eventual failure
of single species due to poor growth or survival (e.g.,
caused by interspecific competition, disease, or herbivore attack). In communities with only one or a few
species, phenological development of the stand is relatively synchronous, potentially leading to vegetation
gaps during the growing season (McNaughton 1993).
This lack of species redundancy in low-diversity communities (Levin 1997) will have subsequent effects on
several different ecosystem processes; e.g., it can cause
‘‘open canopies,’’ which differ in canopy structure and
light climate from fully covered stands (Werger and
Hirose 1988). We therefore performed analyses of covariance to control for diversity-induced differences in
plant cover and to examine the strength of additional
effects of diversity on cover-related aboveground eco-
39
system processes (three-dimensional space, light and
nutrient use, and biomass production).
Caveats
A few caveats should be kept in mind when interpreting our results. (1) The implications of our work
for the consequences of species loss depend on the
implied order of species loss which in our case is determined by our constrained random selection of species (see Methods: Experimental design, below). (2)
We simulate species loss by assembling different diversity communities. An alternative approach to evaluate biodiversity–ecosystem functioning relationships
is species removal from existing communities to create
a diversity gradient (Wardle et al. 1999, Diaz et al.
2003). (3) The results of biodiversity experiments are
not directly comparable with observational studies
comparing productivity of natural or semi-natural ecosystems differing in species richness. This is because
biodiversity experiments look at within-site effects of
diversity on ecosystem processes, where environmental
factors are relatively constant. Observational studies
compare across sites that might vary in environmental
factors that could overwhelm the effect of biodiversity.
For the most recent literature surveys on observational
studies see Grace (1999) and Mittelbach et al. (2001).
For further discussion see Hector et al. (2000b, 2001a,
b) and Schmid et al. (2002a).
One of the main aims of the present study is to present a comprehensive analysis of biodiversity–ecosystem functioning relationships that allows us to extend
and, where possible, generalize previous findings to
multiple processes across different sites in temperate
and Mediterranean grassland ecosystems and to see
how these ecosystem effects of biodiversity develop
over time. The inclusion of sites with contrasting soils
and climate, and with different regional sets of taxa,
should add robustness to our test. We report multisite
analyses of diversity effects on 11 ecosystem process
variables: (1) above- and (2) belowground biomass
over the full three years of the BIODEPTH experiment,
(3) community cover, (4) stand height, (5) canopy biomass density, (6) center of gravity, (7) light use, (8)
nitrogen pools in aboveground vegetation, (9) available
inorganic nitrogen in soil, and (10, 11) decomposition
of two different standard materials in the third year of
the experiment. We ask the following specific questions: (1) How do different ecosystem processes vary
in their response to altered plant diversity? (2) How do
the effects of the manipulation of plant diversity on
these processes develop over time? (3) Does this extended analysis of the ecosystem process variables over
three years present a picture of consistent biodiversity
effects across our eight sites or of location-specific differences?
METHODS
Study sites
This study was carried out at eight experimental sites
along north–south and east–west transects across Eu-
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
40
TABLE 1. Comparison of ecosystem properties, by site, in the experimental (Exp.) plots at the highest diversity level with
unmanipulated reference (Ref.) plots established in adjacent grasslands.
Ecosystem property
Plot†
Aboveground biomass
(g/m2)
Canopy height
(cm)
1,051.8 6 84.7
615.5 6 135.8\
72.1 6 2.9
58.0 6 2.8
78.8 6 10.8
75.0 6 13.2
423.3 6 102.5
400.0 6 142.6
ND
ND
18.0
21.0 6 6.0¶
99.4 6 0.6
97.5 6 2.5
232.7 6 50.7
271.6 6 153.1
53.4 6 4.4
43.3 6 0.0
8.0
11.0 6 0.7
98.1 6 1.0
96.5 6 0.9
11.0
14.8 6 1.0¶
99.0 6 0.7
ND
Type
No.
Species richness‡
(no./plot)
Germany
Exp.
Ref.
6
5
16.0
16.0 6 0.0\
Portugal
Exp.
Ref.
4
4
14.0
14.5 6 1.7
Greece
Exp.
Ref.
8
2
Ireland
Exp.
Ref.
10
4
UK: Silwood
Exp.
10
Ref.
5
Vegetation cover
(%)
93.2 6 1.9
100.0 6 0.0
1065.5 6 53.2
951.0 6 88.7
559.2 6 53.8
680.5 6 119.5
37.0 6 1.9
38.0 6 2.3
29.5 6 1.1
27.5 6 1.9
Notes: Data are means 6 1 SE; ND 5 not determined. Data are from year 3 of the experiment, except those in italics are
from year 2.
† Reference plots were only available at five sites.
‡ Species richness of experimental plots is number of species sown.
§ PAR: photosynthetically active radiation.
\ Only two reference plots were sampled.
¶ Species richness was not determined in the entire plot, but only in the biomass harvest quadrats.
rope that span several thousands of kilometers, in Germany, Greece, Ireland, Portugal, Sweden, Switzerland,
and two sites in the United Kingdom (UK; Silwood
Park near London, and Sheffield). Sites differed widely
in climate, soil conditions, and other major environmental factors (Hector et al. 1999, Joshi et al. 2001).
For more detailed information on single sites, for Switzerland see Diemer et al. (1997), Joshi et al. (2000),
Koricheva et al. (2000), Spehn et al. (2000a, b), Stephan et al. (2000), Diemer and Schmid (2001), Pfisterer
and Schmid (2002), and Pfisterer et al. (2004); for Sweden see Mulder et al. (1999, 2002), Koricheva et al.
(2000), and Jumpponen et al. (2002); for Germany see
Gastine et al. (2003) and Scherer-Lorenzen et al.
(2003); for Greece see Troumbis et al. (2000, 2002);
for Portugal see Caldeira et al. (2001); and for Silwood
see Hector et al. (2000a, 2001a).
Species selection
At each study site, we compiled a list of species
locally present in the grasslands. The communities
therefore consisted of species that were typical of our
sites and that frequently co-occur on the plot scale. We
did not use atypical species. At each site the top level
of diversity aimed to approximate average levels of
diversity found in quadrats of plot size in the local
grasslands (see Table 1).
spring 1996 at all other sites. Plots of 2 3 2 m (2 3
5 m in Sweden) were seeded with 2000 viable seeds/
m2 divided equally between the number of species in
each experimental plant community (Hector et al.
1999). Plant species were representative of local grassland communities and seeds were locally collected as
far as possible, or otherwise purchased from national
commercial sources providing regional seeds. Reciprocal transplant tests with three common species revealed strong patterns of adaptations of local varieties
to their ‘‘home’’ sites (Joshi et al. 2001). Prior to sowing, the existing vegetation of the study sites was removed and the soil seed bank was eliminated by continuous weeding (Switzerland, Sweden), steam sterilization (Germany), heat (soil was covered with black
plastic for 2.5 months, Portugal), methyl bromide application (Ireland, Greece, UK: Silwood) or by establishing plots on calcareous sterilized sand with added
nutrients (UK: Sheffield). To reduce post-application
effects of methyl bromide on legumes or to ensure that
even a rare legume would be inoculated (Sweden), Rhizobium was applied. Plots were weeded to remove unwanted species emerging from the remaining seed bank
or invading from outside the experimental site. The
plots were not subsequently fertilized during the experimental period. Former land use of the sites is listed
in the Supplement (Table S1).
Establishment of the experimental communities
Soil conditions
The field experiments were established in spring
1995 in Switzerland, autumn 1996 in Portugal, and
Initial soil conditions (after the preparation of the
field sites) were measured either in each plot or, where
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
TABLE 1.
41
Extended.
PAR transmittance§
(%)
Belowground biomass
(g/m2)
Soil inorganic N
(mg/kg)
4.33 6 1.55
19.91 6 1.40
363.1 6 114.0
869.0 6 320.7\
1.88 6 0.50
3.61 6 0.46
0.72 6 0.08
0.66 6 0.23\
ND
ND
ND
ND
77.5 6 16.1
ND
3.00 6 0.11
ND
0.02 6 0.00
ND
0.02 6 0.00
ND
38.94 6 2.05
49.62 6 18.27
554.1 6 103.7
451.8 6 205.4
1.80 6 0.11
ND
1.32 6 0.01
1.52 6 0.00
0.07 6 0.02
0.08 6 0.05
3.06 6 0.32
5.56 6 1.60
689.4 6 46.8
808.6 6 76.1
8.69 6 1.52
ND
0.59 6 0.02
0.58 6 0.05
ND
ND
5.78 6 0.97
1.47 6 0.4
1522.3 6 311.4
1905.0 6 673.4
1.36 6 0.16
ND
0.36 6 0.01
0.30 6 0.02
0.10 6 0.01
0.04 6 0.02
plots had not yet been established, by using a stratified
random-sampling scheme in which the experimental
area was divided into a grid of at least 10 cells per
block. Soil cores (4-cm diameter 3 20 cm deep) were
taken at random within each plot or cell, mixed, and a
subsample taken to assess soil pH, available phosphorous, total carbon, total nitrogen, ammonium, and nitrate. At some sites, information on other nutrients was
also collected, depending on the methods of local laboratories. To allow comparison between sites, one composite sample from each block at each site was analyzed
for total C, total N, available P, and pH (in CaCl2) at
the Institute of Environmental Sciences, University of
Zurich (Zurich, Switzerland), following standard methods (Anonymous 1995; see Supplement: Table S1).
Experimental design
Our experiment has a multilevel, hierarchical design
that tests the effects of diversity in the context of differences between sites and due to species composition.
The experiment focuses on species number by establishing a gradient of species richness at each site, with
five levels of diversity ranging from monocultures to
higher diversity mixtures approximately matching
background levels of diversity in comparable semi-natural grasslands at each site (Table 1; Supplement: Table
1). Species mixtures were assembled by random draws
from the local pool of typical co-occurring species (see
Species selection, above). Species selection at each site
was constrained to vary the number of functional
groups—grasses, N-fixing legumes, and other herbaceous dicots (forbs)—within the different levels of species richness and so that all communities included
grasses (with the exception of a few polycultures deliberately established at the Swedish sites to enable
some test of the effects of the grass functional group).
Cotton decomposition Wood decomposition
(% per day)
(% per day)
At low levels (one or two species), only one or two
functional groups could be included, whereas at high
levels it was unrealistic that all species belonged to
only one or two functional groups. As a result the design unavoidably contains empty cells within the species richness and functional-group richness design matrix, and these two variables are therefore partially positively correlated (Schmid et al. 2002a). Each particular
combination of species richness and number of functional groups (diversity level) was replicated with different species compositions (called ‘‘assemblages’’ in
Hector et al. [1999] and in the Supplement to indicate
inclusion of both monocultures and mixtures; here we
use ‘‘composition’’ to be explicit about what this treatment replicates) at each site to separate the effects of
species richness from the effects of species composition (Givnish 1994, Tilman et al. 1997b, Allison 1999,
Schmid et al. 2002a, b). The number of replicates within diversity levels was reduced with increasing species
richness because it was anticipated that the variability
of ecosystem processes would be lower at high diversity as the species overlap among replicates inevitably
increases if the total species pool is fixed (Schmid et
al. 2002a, b). Unlike most other biodiversity experiments, BIODEPTH was designed to also quantify the
variation due to species composition (variation left in
the residual error of other analyses). Every monoculture and polyculture was repeated (with random positioning) in two replicated blocks per site (except Portugal, with a fully randomized design without blocks).
There were two replicates of all compositions, except
there were four replicates for one composition in Portugal, two in Ireland, one in Sweden, and one in UK:
Sheffield. In total, the experiment comprised 480 plots,
which replicated 200 different species compositions
spread across eight sites. Some compositions (16
E. M. SPEHN ET AL.
42
monocultures and one two-species mixture) occurred
at more than one site, mainly at two to four sites, with
two monocultures (Plantago lanceolata, Trifolium repens) occurring at six of the eight sites. This overall
hierarchical design contains multiple error terms,
which are explained under Statistical analysis, below.
Experimental vs. ‘‘natural’’ reference grasslands
In addition, at some sites (Germany [G], Portugal
[P], Greece [GR], Ireland [IR], and UK Silwood) we
established unmanipulated reference plots in neighboring grasslands to provide a natural comparison for
the variables measured in our experimental plots. In
general, all ecosystem properties measured in the unmanipulated reference plots of adjacent grasslands at
5 of the 8 sites (G, P, GR, IR, and UK Silwood) were
very similar to the values obtained in the highest diversity level of the experimental plots (Table 1). Species richness matched the sown species numbers of the
highest diversity levels of the experimental plots very
closely, thus confirming the experimental design that
adjusted the highest diversities to the background level
of comparable grasslands. Some differences were observed at the German site for measures related to or
influencing productivity (aboveground and belowground biomass, canopy height, photosynthetically active radiation transmittance, soil mineral N—but not
cover) with higher values for the experimental plots,
due to higher nutrient availability at the experimental
site (Scherer-Lorenzen 1999). In Greece the experimental communities had a taller canopy than the reference plots, whereas in UK Silwood the experimental
plots had higher PAR transmittance and somewhat lower biomass values than the reference.
Realized species richness
Realized plant-species richness was assessed from
the biomass samples taken at harvest and the visual
cover estimates at the whole plot level taken over the
entire growing season. From all the surveys, we compiled a list of all the plant species present in each plot
and year.
Plant biomass
Aboveground biomass of plants was determined by
harvesting standing crop above 5 cm in one or two
permanent quadrats of 20 3 50 cm per plot once
(Greece, Portugal, Sweden, UK) or twice (Switzerland,
Germany, Ireland) a year. Afterwards, the entire plots
were mown to 5 cm, following the traditional haymeadow managements. Plant samples were oven dried
to a constant dry mass (constancy) and weighed.
Canopy structure and light use
The total cover of vegetation, i.e., the percentage of
ground area covered by live plants, and the ground area
covered by dead plants and litter, was estimated visually before each main harvest. The average canopy
Ecological Monographs
Vol. 75, No. 1
height was measured with a sward stick (Hill Farming
Research Organisation [HFRO], Edinburgh, UK) at the
time of the biomass harvests. A clear window of 2 3
1 cm was lowered vertically down the stick towards
the canopy (Barthram 1986) and the height of first contact with the canopy recorded, and the plot average of
at least 10 repeat measurements was taken. At the time
of peak biomass each year, aboveground biomass was
harvested in canopy layers in a 20 3 50 cm permanent
quadrat. Plants were cut at least at 5 cm, 20 cm, 35
cm, and 50 cm above the ground level (with a few
exceptions; Ireland, no strata at 35 cm; Switzerland,
additional strata at 15 cm). The number of strata depended on total canopy height (for Greece, Portugal,
Sweden, UK Silwood, and Switzerland, last strata .50
cm; for UK Sheffield, last strata .35 cm; for Germany,
strata every 15 cm to the top of each individual canopy). Samples were dried to constancy and weighed.
We then calculated biomass density per layer as a measure of 3-dimensional space filling (see Naeem et al.
1994, 1995) by dividing the height of each layer by
biomass (biomass/volume of layer, in grams per cubic
meter). To get a measure of vertical biomass-density
distribution in the canopy, we calculated the height of
the center of gravity by multiplying the biomass of each
layer with the mean height of the layer (5z) and dividing the sum of z by the total biomass (Spehn et al.
2000b).
Photosynthetically active photon-flux density (PPFD)
was measured with a ceptometer (Delta-T Sunscan ceptometer; Delta-T Devices, Cambridge, UK) or LI-COR
Line Quantum Sensor (Lincoln, Nebraska, USA) at the
base of the canopy (3–5 cm above soil) in each plot
before the main biomass harvest was taken. We then
calculated the mean of at least three individual measurements per plot.
Nitrogen in aboveground biomass
Dried aboveground biomass samples were ground
and analyzed for nitrogen content. Nitrogen in percentage of dry mass was measured by dry combustion
using an automated C-H-N analyzer at three sites (Switzerland, LECO CHN-900 [LECO Corporation, Saint
Joseph, Michigan, USA] Germany, C/N analyzer, Carlo
Erba NA 1500, [Carlo Erba, Mailand, Italy]; Sweden,
Europa Scientific ANCA-NT). In Ireland, Greece, Portugal, and UK Silwood the samples were digested and
analyzed with a semi-automatic Kjeldahl procedure
(Tecator Herndon, Virginia, USA).
Root biomass and length
We measured total community root biomass by taking two soil cores from each plot, avoiding the central
area containing the permanent sampling area of the
aboveground biomass harvests. Soil cores (4-cm diameter) were divided into two or four strata (0–10 cm
and 10–20 cm or every 5 cm) and roots were extracted
by washing and sieving (1-mm mesh size). Samples
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
were dried until constancy and weighted. At two sites
(Switzerland, Germany), fine roots (,1 mm) were separated and length was determined by an image analysis
system (Delta-T area meter [Delta-T Devices, Cambridge, UK]).
Decomposition
We measured the dry mass loss of cotton per day in
the topsoil as an indicator for short-term decomposition. The cotton was a standard organic substrate containing ;95% cellulose, with an initial N concentration
of 0.09%. Four repeat strips of cotton (12.2 3 12.5
cm—identical source used at all sites) were buried vertically (0–10 cm) in each plot over several months
during the main growing period. Cotton strips were
protected against rodents and other physical damage
with a 1-mm nylon mesh (Germany, Greece, Ireland,
UK Silwood, Sweden, Switzerland). Long-term decomposition was measured by dry mass loss of wooden
birch sticks. Standardized sets of three flat birch sticks
(11.5 3 0.9 3 0.2 cm) were bundled using polyester
thread, and buried vertically 1 cm below the surface,
once or twice per growing season. Four bundles of
sticks were used per plot and mean estimates of yearly
decomposition were obtained from differences in dry
mass. Only the central stick in the bundle was monitored. Initial N concentration of the sticks was 0.08%.
Soil nitrogen
Soil-soluble or exchangeable N concentrations were
determined according to a generally agreed sampling
protocol, but with different numbers of samples per
year and different methods of chemical analysis across
sites. Samples were taken with soil cores (diameter
range: 2–5 cm) within the main rooting depth (10–20
cm, depending on site conditions). Two to 10 samples
per plot were taken and pooled for the analysis. From
the sieved samples (2-mm mesh size), aliquots were
extracted either with KCl solution (Switzerland, Germany, Greece, and UK Silwood), or H2SO4 (Kjeldahlprocedure, in Portugal and Ireland), and ammonia and
nitrate concentrations were determined using standard
soil-laboratory procedures. A second subsample was
dried and the water content was used to calculate total
plant–available mineral nitrogen (Nmin) as NH41-N plus
NO32-N milligrams per kilogram of dry mass. Here,
we present the results of the third year of the experiment. The soil was not sampled at the same time across
the sites, so in our analysis we had to combine measurements dating early in the season (Ireland and Portugal), dating mid-season (Switzerland, Germany,
Greece), and from the end of the growing season (UK
Silwood). Soil N was not measured in Sweden and UK
Sheffield during the third year.
Statistical analysis
The data were analyzed using the general-linearmodeling approach to multiple regression (e.g., Neter
43
and Wasserman 1974), implemented in GENSTAT
(Payne et al. 1993), and the results of the model-fitting
sequences were summarized in ANOVA tables (Green
and Tukey 1960, Schmid et al. 2002b). The sequential
analysis determined by our design and a priori hypotheses included main effects of site, block, species
richness, functional group richness, and species composition, as well as interactions of diversity and species
composition with sites. The effects of blocks (nested
within sites) and species composition (nested within
diversity) were considered random factors. Due to the
hierarchical structure of the design, as indicted by the
nesting described above, the experiment had multiple
error terms (Schmid et al. 2002b). The error structure
and hypothesis testing are briefly explained here with
the expected mean squares and error term assignment
given in full in Table 2. To account for spatial variation,
sites were tested against blocks (within sites) and
blocks against plots (within blocks; i.e., the overall
residual). The most important aspect is that our design
explicitly tests species and functional-group richness
effects against species composition effects (and similarly the diversity-by-site interactions are tested against
the composition-by-site interaction). Our design contains terms for the main effects of sites and species
composition plus their interaction. However, it is important to note that to perfectly partition the two main
effects and the site-by-composition interaction would
require a design that repeated the same species composition at all eight sites. This was of course impossible
due to the different plant communities present at the
different locations. Instead, most species compositions
are unique within a single site, with a smaller number
(16 monocultures plus a single two-species mixture)
occurring at more than one site. Consequently, it was
impossible to avoid the site and composition terms being partly confounded and we cannot perfectly partition
the variation due to site, species composition, and their
interaction, and the values presented should be treated
accordingly.
To test the effects related to time, we included year
as a main factor in the analyses of realized species
richness, biomass, and cover (Table 2), using the approach of orthogonal contrasts to repeated-measures
analysis (Elashoff 1986). In this approach, polynomial
contrasts are formed and tested against their own error
terms, avoiding the problem of serial correlation and
therefore the need to adjust the degrees of freedom (see
Rosenthal and Rosnow 1985:65). Studies that conduct
a large number of different tests sometimes employ a
correction for multiple comparisons—often the sequential Bonferroni (Rice 1989)—because when inspecting summaries of a large number of tests, some
significant results will be expected by chance alone.
We do not use such a correction here for several reasons. First, most response variables showed significant
effects, demonstrating that the experiment-wide level
of significance is therefore far higher than would be
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
44
TABLE 2.
Term
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
28
19
20
21
22
Summary of repeated-measures ANOVA of realized species richness per plot from years 1–3.
Source of variation
Site
Block (within site)
Species richness (log2)
Species richness (deviation from log-linear)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species richness (log2)
Site 3 Functional richness (linear)
Composition
Site 3 Composition
Plot
Year
Year 3 Site
Year 3 Block
Year 3 Species richness (log2)
Year 3 Functional richness (linear)
Year 3 Site 3 Species richness (log2)
Year 3 Site 3 Functional richness (linear)
Year 3 Composition
Year 3 Composition 3 Site
Residual
Total
df
7
7
1
9
1
1
7
7
173
28
238
1
7
7
1
1
7
7
183
28
700
1421
SS
937.0
2.1
16 749.3
4126.7
4.8
1.0
111.6
6
67.0
0.0
41.0
0.5
109.7
0.5
0.6
0.9
222.4
28.9
66.0
0.0
191.6
22 668.0
MS
133.9
0.3
16 749.3
458.5
4.8
1.0
15.9
0.9
0.4
0.0
0.2
0.5
15.7
0.1
0.6
0.9
31.8
4.1
0.4
0.0
0.3
F
446.3
1.7
41 873.3
1146.3
12.0
2.5
···
···
···
0.0
0.6
1.7
157.0
0.3
1.5
2.3
1 370 689.7
176 724.1
17 241.4
0.0
† Expected mean squares (without coefficients) show variance components for random and fixed effects and multiple error
terms. Unique variance components of each term are in boldface; s refers to random-effect components, S to fixed-effect
components. Subscript definitions: P 5 Plot, B 5 Block, L 5 Site location, LC 5 Site 3 Composition, S 5 Species richness
(log2) S9 5 Species richness (deviation from log-linearity), F 5 functional richness (linear), F9 5 Functional richness (deviation
from linearity), C 5 Composition, Y 5 Year; YB 5 Year 3 Block, YL 5 Year 3 Site, YLC 5 Year 3 Site 3 Composition,
etc.
‡ The appropriate error for each term is that which contains all other variance components except the one being examined
so that the variance ratio (in this case F) test removes these shared components, revealing the unique effect.
expected by chance alone. Second, our results are based
on strong a priori logical arguments and theory, as well
as supported by other recent studies (see Introduction,
above, and Discussion: Linking diversity effects . . . ,
below). This also allowed us to set our main statistical
model a priori. Third, such techniques increase the rate
of type-II errors (failing to reject the null hypothesis
when there is a significant result). The mathematical,
logical, and practical arguments against applying multiple comparison methods, particularly in situations
like our own, can be found in more detail in Moran
(2003) and references therein.
The role of legumes in generating biodiversity effects observed in our experiment has been addressed
in several previous publications (Scherer-Lorenzen
1999, Jumpponen et al. 2002, Mulder et al. 2002, Spehn
et al. 2002, Scherer-Lorenzen et al. 2003). So here we
restrict ourselves to two main questions. First, does the
presence of legumes explain a significant amount of
the species composition effect? Second, does controlling for the effects of legumes (entering legumes (presence vs. absence) first into the model as covariate)
eliminate the effects of species richness?
Additive partitioning of biodiversity effects
The additive-partitioning method is a generalization
of the relative yield (Harper 1977) and proportional
deviation from expected value (D) approaches (Loreau
1998). Using the method, the effect of biodiversity on
aboveground biomass production can be partitioned
into a selection effect and a complementarity effect
(which sum to the net biodiversity effect). The net biodiversity effect, DY, is measured by the difference between the observed yield of a mixture and its expected
yield based on the average of the monoculture yields
of the component species. The selection effect is the
standard statistical covariance effect, in our case between the monoculture yield of species and their
change in relative yield in the mixture. Positive selection occurs if species with higher-than-average monoculture yields dominate mixtures and negative selection
when species with lower-than-average biomass dominate mixtures. Note that calculation of the selection
effect uses N (number of species) rather than N 2 1 in
the calculation of the covariance term since it is performed on the entire population of species in each mixture rather than a sample of species. The complementarity effect measures the net change in the average
relative yield of species. A positive complementarity
effect occurs when increases in some species are not
exactly compensated by decreases in others, which can
indicate resource-partitioning and related niche-differentiation processes. As the approach requires a comparison between performances of species in mixture
and in monoculture, it can only be applied to the subset
of experimental mixtures that contained species for
which all monoculture yields were available. These
various effects are related in the additive partitioning
as follows:
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
TABLE 2.
Extended.
P
SS
,0.001
0.1
,0.001
,0.001
,0.001
0.116
···
···
···
1.000
1.000
0.197
,0.001
0.934
0.222
0.135
,0.001
,0.001
,0.001
1.000
%
Expected
O RY
O,i
i
Mi 2
O RY
E,i
Mi
i
O DRY M 5 N · DRY · M 1 N cov(DRY, M )
i
MS†
Error term‡
s2 1 s2P 1 s2B 1 S2L
s2 1 s2P 1 s2B
(s2 1 s2P 1 s2LC9 1 S2S)
(s2 1 s2P 1 s2LC9 1 S2S9)
(s2 1 s2P 1 s2LC9 1 S2F)
(s2 1 s2P 1 s2LC9 1 S2F9)
s2 1 s2P 1 s2LC 1 S2LS
s2 1 s2P 1 s2LC 1 S2LF
(s2 1 s2P 1 s2LC9 1 s2C)
2 )
(s2 1 s2P9 1 sLC
9
s2 1 s2P
s2 1 S2Y
s2 1 s2YB 1 S2YL
s2 1 s2YB
s2 1 s2YLC 1 S2YS
s2 1 s2YLC 1 S2YF
s2 1 s2YLC 1 S2YLS
s2 1 s2YLC 1 S2YLF
2 )
(s2 1 s2YLC9 1 sYC
(s2 1 s2YLC9)
s2
4.1
,0.1
73.9
18.2
,0.1
,0.1
0.5
,0.1
0.3
,0.1
0.2
,0.1
0.5
,0.1
,0.1
,0.1
1.0
0.1
0.3
,0.1
0.8
DY 5 YO 2 YE 5
5
45
i
i
where N 5 number of species in the mixture; Mi 5
yield of species i in monoculture; YO,i 5 observed yield
of species i in mixture, and YO 5 SiYO,i 5 total observed
yield of the mixture; RYE,i 5 expected relative yield
of species i in the mixture, which is simply its proportion seeded or planted (1/N), and RYO,i 5 YO,i /Mi 5
observed relative yield of species i in the mixture; YE,i
5 RYE,iMi 5 expected yield of species i in the mixture,
and YE 5 SiYE,i 5 total expected yield of the mixture;
DY 5 YO 2 YE 5 deviation from total expected yield
in the mixture; DRYi 5 RYO,i 2 RYE,i 5 deviation from
expected relative yield of species i in the mixture;
N·DRY·M 5 the complementarity effect, and N
cov(DRY, M) 5 the selection effect.
The additive-partitioning calculations follow Loreau
and Hector (2001) with a few modifications for dealing
with missing species (species were sometimes entirely
absent—see observed species richness—but sometimes
just absent from biomass samples). Species that were
entirely missing occurred particularly in Portugal and
in the third year of the experiment following a period
late in the second year and early in the third year that
was dry and extremely frosty. Species could have zero
harvested biomass in monoculture, polyculture, or
both. Where a species was missing in monoculture,
expected values could not be calculated. These species
therefore had to be ignored entirely by relative-yieldbased methods (species richness was taken as the original richness minus the number of missing species).
Where a species was present in monoculture but missing from a mixture in which it was established once,
2
11
9
9
9
9
10
10
10
11
21
21
14
21
19
19
20
20
20
21
it contributed the negative value of its expected contribution (i.e., a species expected to contribute 10 g
that was missing contributed 210 g). We standardized
the data used here to make them as comparable across
sites and across years within sites as possible. Details
of the calculations and individual species biomass data
used differ slightly from the single-year data used in
Loreau and Hector (2001), although the results and
conclusions of the analysis of the biodiversity effects
over three years remain similar.
Note that the complementarity effect, and the relative
yields it is based on, cannot distinguish facilitation
from resource partitioning (and related niche-differentiation effects through natural enemies, etc.). Therefore, our complementarity effect covers all of these
effects. More detailed mechanistic approaches are required to identify the underlying biology; we refer to
relevant individual site papers where appropriate.
Presentation of results
Some core ecosystem processes were monitored over
three years and for these we present tables of repeatedmeasures ANOVA. The analyses often produced significant interactions of diversity with site or year or
both. In the figures, we therefore present sites separately, showing observed means for each diversity level
in each year. For some variables, we did not have a
comprehensive data set across all sites and years and
we therefore analyzed only the third year. The figures
for single-year analyses also show sites separately but
provide greater detail: we show the overall regression
slope with means and standard errors of the replicate
pairs of each species composition. We distinguish compositions with legumes from those without (since many
E. M. SPEHN ET AL.
46
Ecological Monographs
Vol. 75, No. 1
FIG. 1. Realized species richness of all sites in the first three years of the experiment. (Realized species richness is the
number of species present in each plot and year.) Large crosses indicate planned species richness (the number of species
sown in the plots at the start of the experiment). Data are means 6 1 SE of all compositions within species richness levels.
of the single-year variables were strongly influenced
by this aspect of diversity).
There was a strong correlation between the planned
and realized diversity gradients over the three years of
the experiment (R2 5 0.93; Table 2). However, realized
numbers of species deviated sometimes from the numbers sown due to two causes: some species failed to
germinate or successfully establish in the first year of
the experiment; and, less commonly, some species went
extinct from plots in the second or third year (Fig. 1).
While many factors had significant effects on realized
richness, they all explained ,1% of the variance apart
from site (4.1%, Table 2). The greatest species loss
occurred in Portugal, partly due to extreme climate
conditions during the third year of the study. Therefore,
because realized richness generally matched planned
richness well, and because planned richness was the
experimental treatment factor, all of the following analyses were calculated with planned richness as explanatory variable. Analyses that replace planned richness
with realized richness show no difference from the general outcome and planned richness actually has significantly more explanatory power than realized richness
(for a discussion of why, see Hector et al. [2000b,
2002b]).
and Sweden. In Ireland, there was an especially strong
positive establishment effect of biodiversity, with low
cover during the first year, but full cover in the subsequent years. Overall, vegetation cover was positively
related to increasing numbers of plant species and functional groups (Table 3). The effect of diversity on cover
was often stronger later in the experiment (except UK:
Sheffield and Ireland). Interestingly, this was not due
to higher increases in cover over time in high-diversity
plots but the opposite: cover in low-diversity plots in
Switzerland, Germany, Sweden, and UK: Silwood decreased from the first to the third year. The southernmost and northernmost European sites in Portugal and
Sweden had lower vegetation cover than the other sites
with less extreme climatic conditions. In these systems
where productivity is limited, measurements in unmanipulated ‘‘reference’’ plots showed that cover was
normally lower (Table 1). In Portugal, a dry season
followed by uncharacteristic frosts killed many plants
in the experimental and reference plots, reducing cover
in both although more strongly in the experimental
plots (average reduction in percent cover from year 2
to 3: experimental plot 5 283%, reference plots 5
256%). In Sweden, ice cover created bare patches in
one of the two blocks in the first and second winter
and the most limiting factor for growth was low winter
temperature, resulting in short growing seasons due to
long snow cover or deep soil frost.
Vegetation cover
Aboveground biomass
The effects of plant diversity (species and functional
richness and species composition) on vegetation cover
varied strongly with year and site (Fig. 2a, Table 3).
Absolute cover increased over the three years of the
experiment at five sites but not in Germany, Portugal,
Overall diversity had a positive effect on aboveground biomass. Aboveground biomass increased log
linearly with species richness and linearly with increasing number of functional groups (Table 3). However,
across all three years the deviation around the species-
RESULTS
Realized species richness
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
47
FIG. 2. Species richness effects (a) on vegetation cover (%) and (b) on aboveground biomass of all sites in the first three
years of the experiment. Data are means 6 1 SE of all mixtures within species-richness levels.
richness effect became significant as well, and the effects of diversity varied with sites (Fig. 2b, Table 3).
At most sites (except Portugal) the biodiversity–productivity relationship became more positive over time.
However, in Greece the lack of a significant effect of
diversity on productivity persisted. As reported in Hector et al. (1999), not surprisingly production was lower
in the southern and northern sites where growing seasons are more restricted than in the mid-European sites.
Biomass was usually lowest in the first year of the
experiment—except in Portugal where it declined in
year three due to extreme climatic events (see above)—
and then increased with time (Ireland, UK: Sheffield)
or settled at similar levels in years two and three (other
sites; Fig. 2b).
Additive partitioning of biodiversity effects
There was a main effect of species richness on the
complementarity effect: across all sites and years the
effect of species richness on aboveground biomass production was positive with a slope of 2.5 6 0.5; F1,69 5
10.1, P , 0.01; Table 4). The strength of the richness–
complementarity effect relationship was weaker at the
start of the experiment than in the second and third
year (Fig. 3a), except for Portugal. The grand means
for the net, complementarity, and selection effect were
all highly significantly positive by the end of the experiment, and the net and complementarity effect increased on average over the three years (net effect years
1–3: 37 6 6.4 g/m2, 80 6 9.5 g/m2, 99 6 9.4 g/m2,
respectively; complementarity effect: 54 6 11.6 g/m2,
55 6 9.0 g/m2, 77 6 8.8 g/m2, respectively; Fig. 3a),
whereas the selection effect was more variable (selection effect: 217 6 9.9 g/m2, 25 6 6.1 g/m2, 21 6 4.8
g/m2, respectively; Fig. 3b) (all data: means 6 1 SE).
The selection effect was initially negative, meaning
that communities were initially dominated by species
with a lower-than-average monoculture biomass.
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
48
TABLE 3. Results of repeated-measures ANOVA for percent cover of vegetation and aboveground biomass (in grams per
square meter) during the first three years of the experiment.
Cover
Source of variation
df
SS
Site
Block (within site)
Species richness (log2)
Species richness (deviation)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species richness (log2)
Site 3 Functional richness (linear)
Composition
Site 3 Composition
Plot
Year
Year 3 Site
Year 3 Block
Year 3 Species richness (log2)
Year 3 Functional richness (linear)
Year 3 Site 3 Species richness (log2)
Year 3 Site 3 Funct. richness (linear)
Year 3 Composition
Year 3 Composition 3 Site
Residual
Total
7
7
1
9
1
1
7
7
173
28
238
1
7
7
1
1
7
7
183
28
556
1277
288 869
22 884
43 529
6225
6067
117
4697
11 939
139 690
19 188
28 192
18 505
131 094
5424
2550
0.2
5051
709
46 711
8068
58 835
848 343
Canopy structure
Vertical distribution of biomass varied markedly between sites (Fig. 4, Table 5). This variation is partly
due to differences in biomass yields, but also to strong
differences in canopy heights and therefore in how
dense the biomass is packed in the canopy. In some
sites, canopies were tall, most notably in Germany,
leading to a moderate biomass density (;1000 g/m3),
while in other sites canopies were short and dense,
F
P
12.6
27.6
53.9
0.9
7.5
0.1
1.0
2.5
1.2
5.8
1.1
174.9
24.2
7.3
10.0
0.0
2.5
0.4
0.9
2.7
0.002
0.000
0.000
0.565
0.007
0.703
0.466
0.040
0.313
0.000
0.147
0.000
0.000
0.000
0.002
0.978
0.039
0.922
0.691
0.000
MS
41 267.0
3269.0
43 529.0
691.7
6066.6
117.4
671.0
1705.6
807.5
685.3
118.5
18 504.9
18 727.7
774.8
2550.0
0.2
721.6
101.3
255.3
288.2
105.8
SS
%
34.1
2.7
5.1
0.7
0.7
0.0
0.6
1.4
16.5
2.3
3.3
2.2
15.5
0.6
0.3
0.0
0.6
0.1
5.5
1.0
6.9
especially in Ireland (;1500 g/m3) and UK: Sheffield
(;2500–3000 g/m3). Over the three years there was no
significant overall diversity effect; instead, only individual layers at some sites differed significantly in density with changing diversity (Table 5). In contrast, during the third year, both canopy height and center of
gravity of vertical biomass distribution significantly increased with diversity, although the effects varied with
site (Fig. 5a, Table 6). Diversity-related changes in
TABLE 4. Net biodiversity effect (net effect), partitioned into complementarity (complementarity effect) and selection
mechanisms (selection effect) on aboveground biomass for all sites in the third year of the experiment.
Complementarity
effect
Net effect
Source of variation
Grand mean (constant)
Site
Block (within site)
Species richness (log2)
Species richness (deviation)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species rich. (log2)
Site 3 Functional rich. (lin.)
Composition
Plot
Year
Year 3 Site
Year 3 Block
Year 3 Species richness (log2)
Year 3 Functional richness (lin.)
Year 3 Site 3 Species rich. (log2)
Year 3 Site 3 Funct. rich. (lin.)
Year 3 Composition
Residual
Total
df
SS
MS
F
P
1
7
6
1
5
1
1
5
7
69
101
1
7
7
1
1
5
7
75
270
577
33 552.1
5563.5
2150.7
2852.0
642.3
2612.4
184.5
959.6
1475.2
12 378.9
5447.1
3029.6
2568.9
847.4
215.4
235.2
933.9
947.5
5357.5
13 661.9
62 063.7
33 552.1
794.8
358.5
2852.0
128.5
2612.4
184.5
191.9
210.7
179.4
53.9
3029.6
367.0
121.1
215.4
235.2
186.8
135.4
71.4
50.6
187.0
2.2
6.7
15.9
0.7
14.6
1.0
1.1
1.2
3.3
1.1
59.9
3.0
2.4
3.0
3.3
2.6
1.9
1.4
0.0000
0.1756
0.0000
0.0002
0.6136
0.0003
0.3141
0.3847
0.3287
0.0000
0.3398
0.0000
0.0833
0.0217
0.0866
0.0736
0.0311
0.0822
0.0251
SS
%
54.1
9.0
3.5
4.6
1.0
4.2
0.3
1.5
2.4
19.9
8.8
4.9
4.1
1.4
0.3
0.4
1.5
1.5
8.6
22.0
SS
MS
21 501.0 21 501.0
4745.1
677.9
1589.7
265.0
2142.9 2142.9
1224.3
244.9
1174.5 1174.5
177.6
177.6
2112.1
422.4
1487.9
212.6
14 703.8
213.1
6867.4
68.0
1474.0 1474.0
1962.7
280.4
684.0
97.7
0.8
0.8
17.0
17.0
499.8
100.0
1380.3
197.2
6726.5
89.7
18 027.9
66.8
66 998.2
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
TABLE 3.
49
Extended.
Biomass
df
SS
7
7
1
9
1
1
7
7
173
28
238
1
7
7
1
1
7
7
182
28
664
1418
5 3 107
432 168
9 954 045
2 622 586
827 526
82 332
1 952 473
1 166 519
1.9 3 l07
3 167 946
5 388 968
8 871 089
1.3 3 107
215 396
2 078 579
261 487
1 102 119
380 279
6 961 881
1 219 032
1.9 3 l07
1.5 3 108
7 205 361
61 738.29
9 954 045
291 398.4
827 526
82 332
278 924.7
166 645.6
109 382.9
113 140.9
22 642.72
8 871 089
1 795 133
30 770.86
2 078 579
261 487
157 445.6
54 325.57
38 252.09
43 536.86
28 727.09
aboveground biomass and canopy structure also affected light interception and transmittance. Levels of
transmitted PAR (photosynthetically active radiation)
at the base of the canopies generally declined with
increasing diversity (Fig. 5b, Table 6).
Controlling for diversity effects on vegetation cover
We repeated the analysis of aboveground biomass
(Fig. 2a and b, Table 3) adding vegetation cover as a
TABLE 4.
F
P
116.7
3.2
91.0
2.7
7.6
0.8
2.5
1.5
1.0
5.8
0.8
308.8
58.3
1.1
54.3
6.8
3.6
1.2
0.9
1.5
0.000
0.003
0.000
0.006
0.007
0.387
0.042
0.217
0.574
0.000
0.985
0.000
0.000
0.380
0.000
0.010
0.007
0.311
0.855
0.044
MS
SS
%
34.2
0.3
6.7
1.7
0.6
0.0
1.4
0.8
12.9
2.2
3.9
5.9
8.8
0.2
1.4
0.1
0.7
0.2
4.8
0.8
12.6
covariate. Cover did indeed have a highly significant
effect on biomass (F1, 201 5 133.4, P , 0.001) indicating
that diversity effects on aboveground biomass were
related in part to increased cover. Nonetheless, there
was still a highly significant residual effect of species
richness on biomass (F1, 201 5 46.5, P , 0.001), which
was not explained simply by greater levels of vegetation cover. Effects of diversity (species and functional-group richness) on canopy-structure variables
Extended.
Selection effect
Complementarity effect
F
P
100.9
2.6
3.9
10.1
1.2
5.5
0.8
2.0
1.0
3.1
1.0
22.1
2.9
1.5
0.0
0.2
1.1
2.2
1.3
0.0000
0.1364
0.0015
0.0023
0.3432
0.0218
0.3645
0.0921
0.4407
0.0000
0.4462
0.0000
0.0938
0.1804
0.9250
0.6645
0.3600
0.0437
0.0471
SS
%
32.1
7.1
2.4
3.2
1.8
1.8
0.3
3.2
2.2
21.9
10.6
2.2
2.9
1.0
0.0
0.0
0.7
2.1
10.0
26.9
SS
3572.1
9866.3
351.1
416.4
167.0
52.1
0.1
586.2
2184.1
10 813.2
3218.9
1159.8
1908.6
457.2
724.8
365.3
388.5
306.0
7203.7
14 481.9
54 651.0
MS
F
P
3572.1
1409.5
58.5
416.4
33.4
52.1
0.1
117.2
312.0
156.7
31.9
1159.8
272.7
65.3
724.8
365.3
77.7
43.7
96.1
53.6
22.8
24.1
1.8
2.7
0.2
0.3
0.0
0.7
2.0
4.9
0.6
21.6
4.2
1.2
7.5
3.8
0.8
0.5
1.8
0.0000
0.0006
0.0995
0.1077
0.9559
0.5661
0.9845
0.5903
0.0687
0.0000
0.9986
0.0000
0.0395
0.2930
0.0075
0.0549
0.5469
0.8636
0.0004
SS
%
6.5
18.1
0.6
0.8
0.3
0.1
0.0
1.1
4.0
19.8
5.9
2.1
3.5
0.8
1.3
0.7
0.7
0.6
13.2
26.5
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
50
FIG. 3. Complementarity and selection effects as a function of species richness in the first three years of the experiment
at all sites. (a) Contribution of complementarity to over-yielding of aboveground biomass (complementarity effect). (b)
Selection effect on aboveground biomass. Data are means 6 1 SE of all mixtures within species richness levels. Values are
square-root transformed to meet the assumptions of analyses but preserve the original positive and negative signs.
remained significant for light use, canopy height, and
center of gravity of vertical biomass distribution, P ,
0.05).
Nitrogen in aboveground biomass
Total N content in aboveground biomass increased
with diversity (Fig. 6, Table 7) in parallel with the
increase in aboveground biomass (Table 3, Fig. 2a).
Germany, Switzerland, and Sweden showed the strongest increase in biomass and N content, mainly driven
by legume presence (see The role of legumes . . . , below), whereas at the other sites only a weak increase
was measured. Results of the third year were similar
to published results from the second year of the experiment (Spehn et al. 2002).
Root biomass
The effect of species richness on root biomass was
consistent across sites and generally slightly positive
(Fig. 7, Table 8). However, there were also large and
highly significant differences in root biomass between
sites explaining 53% of total variation, which might
not only reflect differences in vegetation type, soil
chemistry, and texture, but also differences in methodology (e.g., efficiency of root extraction from soils
of differing types).
Soil nitrogen
Total inorganic (soluble) nitrogen concentrations differed strongly among sites (Fig. 8), with mean values
ranging from 1.7 mg/kg (Greece) up to 29.0 mg/kg
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
FIG. 4.
Species richness effects on density of aboveground biomass in different canopy layers at all sites.
(Switzerland), reflecting different soil types, former
land use, and soil fertility (see Supplement) and presumably to a much lesser extent also different soilanalysis methods (see Methods: Soil nitrogen, above).
Consequently, up to 79% of the total variation in soil
N in our full statistical model was attributed to the site
term. There was no overall effect of species richness
on N concentrations in the soil, but increasing the number of plant functional groups led to a decrease in soil
TABLE 5.
51
N (Fig. 8, Table 9). Species composition had the largest
effect, explaining more than half of the sum of squares
remaining after eliminating the large site effect.
Decomposition
The amount of standard material decomposed differed mainly between sites, with Portugal and UK
(Sheffield and Silwood) showing a lower cotton decomposition rate than the other sites (less than 0.5%
Results of repeated-measures ANOVA for different layers of canopy density in the third year.
Source of variation
Site
Block (within site)
Species richness (log2)
Species rich. (deviation from log-lin.)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species richness (log2)
Site x Functional richness (lin.)
Composition
Site 3 Composition
Plot
Layers (log2)
Layers 3 Site
Layers 3 Block
Layers 3 Species richness (log2)
Layers 3 Functional richness (lin.)
Layers 3 Location 3 Species richness (log2)
Layers 3 Site 3 F. richness (lin.)
Layers 3 Composition
Layers 3 Site 3 Composition
Residual
Total
df
SS
MS
F
P
7
7
1
9
1
1
7
7
166
28
229
1
7
7
1
1
7
7
176
27
1304
2001
563 714 954
15 414 174
5 337 796
15 227 455
12 339 538
843 247
22 843 110
3 589 047
466 250 781
48 433 417
335 619 338
1 047 493 706
292 270 197
23 088 453
665
6 519 965
44 494 696
2 411 164
478 809 893
29 928 422
645 909 009
4 060 539 028
80 530 708
2 202 025
5 337 796
1 691 939
12 339 538
843 247
3 263 301
512 721
2 808 740
1 729 765
1 465 587
1 047 493 706
41 752 885
3 298 350
665
6 519 965
6 356 385
344 452
2 720 511
1 108 460
495 329
36.6
1.5
1.9
0.6
4.4
0.3
1.9
0.3
1.6
1.1
3.0
2114.7
12.7
6.7
0.0
2.4
5.7
0.3
2.5
2.2
0.000
0.167
0.170
0.794
0.038
0.584
0.110
0.950
0.066
0.392
0.000
0.000
0.002
0.000
0.988
0.123
0.000
0.943
0.004
0.000
SS
%
13.9
0.4
0.1
0.4
0.3
0.0
0.6
0.1
11.5
1.2
8.3
25.8
7.2
0.6
0.0
0.2
1.1
0.1
11.8
0.7
15.9
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
52
FIG. 5. Species richness effects on (a) total canopy height (circles) and center of gravity of vertical biomass distribution
(squares) and on (b) PAR transmittance (percentage of photosynthetically active radiation measured at the bottom of the
canopies) in year 3 of the experiment at all sites. Solid symbols represent compositions containing one or several legume
species; open symbols represent compositions without legumes. Data are means 6 1 SE of each composition; the regression
slope is from the overall statistical model (Table 6).
TABLE 6. Results of ANOVA of canopy height, center of gravity, and photosynthetically active radiation (PAR) at the base
of the canopy in the third year.
Height
Center of gravity
Source of variation
df
SS
MS
F
P
Site
Block (within site)
Species richness (log2)
Species richness (deviation)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species richness (log2)
Site 3 Funct. richness (lin.)
Composition
Site 3 Composition
Residual
Total
6
7
1
8
1
1
6
6
152
26
202
416
107 185
3122
27 281
1805
23
717
8871
645
84 011
7071
17 000
257 731
17 864
446
27 281
226
23
717
1479
107
553
272
84
40.1
5.3
49.3
0.4
0.0
1.3
5.4
0.4
2.0
3.2
0.000
0.000
0.000
0.914
0.839
0.257
0.001
0.876
0.018
0.000
SS
%
41.6
1.2
10.6
0.7
0.0
0.3
3.4
0.3
32.6
2.7
df
SS
MS
7
7
1
9
1
1
7
7
173
28
238
479
35 706
104
2063
958
13
5
851
409
8296
1570
1961
51 936
5101
15
2063
106
13
5
122
58
48
56
8
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
per day, Fig. 9). Again, this difference probably reflects
a mixture of biological differences and differences in
soil type, etc. (see Methods: Soil conditions, above).
On average, species and functional diversity had a positive effect on cotton decomposition, and functional
diversity also had a positive effect on wood decomposition (Table 10). However, the relationship was
clearly driven mainly by the Greek site, which showed
the strongest response to an increase in species richness
(Fig. 9).
not differ between sites, but showed a significant main
effect (F1, 184 5 22.45, P , 0.001). Finally, for canopy
height, legumes did not contribute significantly to the
effect of species composition (F1, 152 5 0.83, P . 0.3).
A significant (all P , 0.05) residual main effect of
species richness remained, when the presence of legumes was added first as a covariate in the analyses:
for aboveground biomass, (F1, 173 5 39.2 vs. F1, 173 5
91.0 in Table 3), for complementarity effect (F1,69 5
3.5 vs. F1,69 5 10.1 in Table 4), for canopy height (F1, 152
5 38.0 vs. F1, 152 5 49.3 in Table 6), for PAR (F1, 134
5 11.3 vs. F1, 134 5 51.6 in Table 6), for belowground
biomass (F1, 173 5 6.8 vs. F1, 173 5 8.8 in Table 8), and
for soil N (F1, 139 5 11.30 vs. F1, 139 5 0.9 in Table 9).
For the selection effect in this modified analysis, species richness interacted significantly with site ( F5, 69 5
1.1, P 5 0.0068 vs. F5, 69 5 0.7 in Table 4).
The overall effect of species composition
There were significant differences among the 200
different species compositions at the eight sites with
regard to all ecosystem processes (significant Composition or Site 3 Composition terms in Tables 1–9)
except for root biomass and wood decomposition. Because we had 200 different compositions formed from
over 100 different species, we concentrated mainly on
the role of nitrogen-fixing species as a group. Individual species effects are being analyzed for publication
in a later paper.
DISCUSSION
Most biodiversity manipulation experiments to date
have been conducted at single sites (but see Emmerson
et al. 2001 and Van der Putten 2000), usually over
relatively short time periods (but see Tilman et al.
2001) and often focusing on a small number of ecosystem processes (but see Hooper and Vitousek 1997).
Here, we analyze data (see Supplement) on a wide
range of ecosystem processes collected with standardized experimental protocols at eight different European grassland sites for a minimum of three years.
The general conclusion of our study is that ecosystem
functioning at our European grassland sites was influenced by all three of our experimental variables.
Not surprisingly, differences between locations and
differences in species composition both explained
substantial and significant amounts of the variation of
most of the ecosystem processes examined. Nevertheless, against this background of geographic variation, and when tested against the variation due to
species composition, all ecosystem processes examined were significantly affected by changes in one or
more aspects of diversity (species and/or functional-
The role of legumes in generating biodiversity effects
In general, a significant part of the differences between species compositions (after having included the
effect of species and functional-group richness in the
models) was explained by the presence of legumes.
However, there were several significant (P , 0.05) site
3 legume interactions for PAR ( F6,26 5 3.6), belowground biomass (F7,28 5 3.2), and soil N (F6,12 5 18.91).
Similarly, the year 3 site 3 legume interaction was
also sometimes significant in the repeated-measures
analysis: aboveground biomass (F7,28 5 5.8), complementarity effect (F7,75 5 3.5). The effect of legumes
clearly often varied with site, time, or both. However,
sometimes the effects of legumes were simpler. For the
selection effect, the influence of legumes interacted
with time but not site (legume 3 year interaction F1,75
5 11.6, P , 0.01). Second, the positive effect of legumes on cellulose (cotton-strip) decomposition did
TABLE 6.
53
Extended.
PAR
Center of gravity
F
P
343.3
1.8
43.0
2.2
0.3
0.1
2.2
1.0
0.9
6.8
0.000
0.087
0.000
0.023
0.603
0.747
0.069
0.425
0.733
0.000
SS
%
68.8
0.2
4.0
1.8
0.0
0.0
1.6
0.8
16.0
3.0
df
SS
MS
F
P
5
6
1
7
1
1
5
5
134
18
158
341
72 847
4089
21 542
969
943
22
3676
2245
55 906
8511
11 213
181 963
14 569
682
21 542
138
943
22
735
449
417
473
71
21.4
9.6
51.6
0.3
2.3
0.1
1.6
0.9
0.9
6.7
0.001
0.000
0.000
0.938
0.135
0.819
0.223
0.474
0.673
0.000
SS
%
40.0
2.2
11.8
0.5
0.5
0.0
2.0
1.2
30.7
4.7
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
54
FIG. 6. Species richness effects on aboveground biomass nitrogen in year 3 of the experiment. Solid symbols represent
compositions containing one or several legume species; open symbols represent compositions without legumes. Data are
means 6 1 SE of each composition; the regression slope is from the overall statistical model (Table 7).
group richness). Communities with higher levels of
diversity reached higher cover, were often more productive, and utilized resources more completely including light, soil nitrogen and space. In short, our
extended analysis reveals that the effects of diversity
are more widespread than just the previously analyzed
variables and years (Hector et al. 1999, Spehn et al.
2002).
However, the effects of diversity varied with both
site and time. Some processes, such as short-term (cotton) decomposition and canopy spacefilling (biomass
density) were more strongly affected at some sites than
at others, and long-term (wood) decomposition was not
significantly affected by changes in species richness at
any site, but only by the number of functional groups.
In general, the effects of diversity were stronger and
more consistent aboveground than below (see the following section and Joshi et al. 2004).
TABLE 7.
Linking diversity effects to different
ecosystem processes
Aboveground, higher-diversity communities showed
a consistent pattern of higher complementarity-effect
values, greater resource use in terms of light and space,
higher productivity, and therefore larger pools of N in
biomass. This linkage of greater complementarity, resource use, and productivity is consistent with theories
based on niche differentiation and resource partitioning
(Tilman et al. 1997b, Loreau 1998, 2000), but is partly
also based on facilitation (e.g., through fixed N as well
as possible influence of nurse-plant effects, reduction
of physical stress, mechanical support, etc. [Callaway
1995, Mulder et al. 2001, Bruno et al. 2003]). While
interactions between legumes and non-legumes have
been addressed in some ways by theory (Schwinning
and Parsons 1996), the increase in the resource supply
Results of ANOVA for aboveground biomass nitrogen pool in the third year.
Source of variation
df
Site
5
Block (within site)
5
1
Species richness (log2)
Species rich. (deviation from log-linear)
8
Functional richness (linear)
1
Functional richness (deviation)
1
5
Site 3 Species richness (log2)
5
Site 3 Functional richness (linear)
Composition
126
16
Site 3 Composition
Residual
166
Total
339
SS
MS
F
P
1609.7
114.5
736.6
227.3
233.0
2.6
442.2
146.2
2191.5
267.8
963.0
6934.6
321.9
22.9
736.6
28.4
233.0
2.6
88.4
29.2
17.4
16.7
5.8
14.1
3.9
42.5
1.6
13.4
0.1
5.3
1.7
1.0
2.9
0.006
0.002
0.000
0.122
0.000
0.698
0.005
0.181
0.498
0.000
SS
%
23.2
1.7
10.6
3.3
3.4
0.0
6.4
2.1
31.6
3.9
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
55
FIG. 7. Species richness effects on root biomass of the top 20-cm soil layer in year 3 of the experiment. Solid symbols
represent compositions containing one or several legume species; open symbols represent compositions without legumes.
Data are means 6 1 SE of each composition; the regression slope is from the overall statistical model (Table 8).
driven by some species through N2 fixation and how
this affects the relationship between biodiversity and
ecosystem functioning is something that has yet to be
properly addressed by theory.
The picture belowground was not so clear. Existing
theory predicts (Tilman et al. 1997b, Loreau 1998) that
with increasing diversity and therefore higher uptake,
levels of unconsumed soil N should fall due to interspecific differences in exploiting different portions of
the available N pool either in space or in time. This
type of effect has been observed in some other biodiversity experiments (Ewel et al. 1991, Tilman et al.
1996, 1997a, Hooper and Vitousek 1997, Niklaus et
al. 2001; but see Naeem et al. 1995, Hooper and Vitousek 1998, Symstad et al. 1998, Kenkel et al. 2000).
Apart from differences in methods that partly account
for the observed variability between sites, differences
in soil N pools may also be a result of regionally different contributions of biological N2 fixation to overall
N availability. Using isotope techniques at three sites
(Germany, Sweden, and Portugal) we have shown that
TABLE 8.
legumes were actively fixing atmospheric N, which was
also transferred to neighboring non-legume species
(Mulder et al. 2002, Spehn et al. 2002). At all other
sites, except Greece, both biomass and tissue-N data
also support this observation. Depending on the presence or absence of actively N2-fixing legumes in mixtures, we saw an altered pattern of diversity on soil N
pools. Communities without legumes showed little
overall pattern in available soil N. With legumes, however, there was generally a decline in available soil N
with increasing diversity. More detailed analyses revealed that there was no correlation between the abundance of legumes and soil mineral N across all sites,
except UK Silwood where higher legume biomass was
associated with higher soil N pools by the third year.
Lower levels of unconsumed soil N are often associated
with greater root biomass (Tilman et al. 2002, SchererLorenzen et al. 2003). We also observed in communities containing legumes an increase in fine-root biomass with greater diversity, which may imply a higher
uptake capacity for soil nutrients. These results support
Summary of ANOVA for root biomass (0 to 220 cm) in the third year.
Source of variation
df
SS
MS
F
P
Site
Block (within site)
Species richness (log2)
Species rich. (deviation from log-linear)
Functional richness (linear)
Functional richness (deviation)
Site 3 Species richness (log2)
Site 3 Functional richness (linear)
Composition
Site 3 Composition
Residual
Total
7
7
1
9
1
1
7
7
173
28
238
479
68 695 652
1 190 714
1 232 544
1 210 921
400 968
13 811
1 307 461
2 550 668
24 300 029
3 729 102
25 054 230
129 686 100
9 813 664
170 102
1 232 544
134 546
400 968
13 811
186 780
364 381
140 462
133 182
105 270
57.7
1.6
8.8
1.0
2.9
0.1
1.4
2.7
1.1
1.3
0.000
0.132
0.003
0.477
0.093
0.754
0.243
0.027
0.455
0.176
SS
%
53.0
0.9
1.0
0.9
0.3
0.0
1.0
2.0
18.7
2.9
Ecological Monographs
Vol. 75, No. 1
E. M. SPEHN ET AL.
56
FIG. 8. Species richness effects on soil inorganic nitrogen concentrations within the rooting zone (top 20-cm soil layer)
in year 3 of the experiment. Note the different scales for the y-axes. Solid symbols represent compositions containing one
or several legume species; open symbols represent compositions without legumes. Data are means 6 1 SE of each composition;
the regression slope is from the overall statistical model (Table 9).
the view that more diverse systems are more efficient
in nitrogen uptake when additional N input occurs
through fixation or when soils are not N limited (Scherer-Lorenzen et al. 2003).
The belowground process that showed the least influence of diversity was decomposition. There was a
significant effect only of functional-group richness and
not of species richness on wood decomposition. The
increase in cotton decomposition with greater diversity
was clearly mainly driven by a strong relationship only
at the Greek site. This is a good example of different
processes responding differently to changes in diversity, since in Greece, where plants grew on an unusual
soil high in Mg21, Ca21 and very low in PO432 (see
Supplement), aboveground production and many other
processes were largely unaffected (Troumbis et al.
2000). In general, direct plant-diversity effects on processes such as wood decomposition would imply niche
TABLE 9.
differentiation between plant species influencing this
particular process (Lawton 2000). More plausible are
indirectly driven diversity effects through changes in
biotic (e.g., fauna) and abiotic (e.g., moisture) soil conditions, which are more likely to be idiosyncratically
linked to species identity than to species richness, as
shown by earlier work on diversity effects on litter
decomposition (Wardle et al. 1997).
Why do such differences in above- and belowground
responses occur? First, belowground responses may be
delayed due to long-term effects of previous land use—
and in our case site preparation (see Methods: Establishment of experimental . . . , above)—on soil chemical, physical, and biological properties. This could
mask weak diversity–function relations for some time
until they manifest over the long-term (Nilsson et al.
1999, Compton and Boone 2000, Hooper et al. 2000).
Second, indirect effects of plant diversity on microbial
Results for ANOVA for soil mineral nitrogen pools in the third year.
Source
df
Site
5
Block (within site)
5
1
Species richness (log2)
Species rich. (deviation from loglinear)
8
Functional richness (linear)
1
Functional richness (deviation)
1
5
Site 3 Species richness (log2)
5
Site 3 Functional richness (linear)
Composition
139
15
Site 3 Composition
Residual
181
Total
366
† After eliminating the large site effect.
SS
MS
F
P
36 329
249
35
7265.8
50.0
35.0
145.9
4.9
0.9
0.000
0.000
0.349
79.0
0.5
0.1
2.6
0.4
130
289
10
42
614
5501
960
1845
46 004
16.3
289.0
10.0
8.4
122.8
39.6
64.0
10.2
0.4
7.3
0.3
0.1
1.9
0.6
6.3
0.913
0.008
0.616
0.983
0.151
0.924
0.000
0.3
0.6
0.0
0.1
1.3
12.0
2.1
4.0
100.0
1.3
3.0
0.1
3.0
8.3
52.4
9.9
19.1
100.0
SS
%
SS
%†
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
February 2005
57
FIG. 9. Species richness effects on decomposition of cotton-strip standard material (percentage dry-mass loss per day)
in year 3 of the experiment. Solid symbols represent compositions containing one or several legume species; open symbols
represent compositions without legumes. Data are means 6 1 SE of each composition; the regression slope is from the overall
statistical model (Table 10).
processes (e.g., via altered soil microclimate or via
changes in root exudates) such as mineralization, nitrification, or immobilization may equal or even exceed
direct effects of plant uptake, as shown by Hooper and
Vitousek (1998), Niklaus et al. (2001), and Zak et al.
(2003). Third, soil food-web properties and related soil
processes like decomposition and mineralization might
be more tightly coupled to soil abiotic conditions and
species composition than to plant species diversity
(Wardle et al. 1999), which may explain why no consistent pattern was found over our large gradient in site
characteristics. Fourth, apart from abiotic stress conditions at particular sites, functional redundancy of soil
organisms (e.g., Brussaard et al. 1997, Bradford et al.
2002), and the fact that belowground processes are
much slower than aboveground responses (Wardle et
al. 1999, Wardle 2002) have been suggested as explanations for the lack of quick responses in biogeochemTABLE 10.
ical processes to changes in producer diversity (Joshi
et al. 2004).
Variation of diversity effects over time
For the variables measured over multiple years, biodiversity effects were often lowest in the first year and
increased in year two, or three, or both. For example,
aboveground biomass increased in its mean value (Ireland and UK Silwood) or in diversity slope (Germany,
Switzerland, Sweden, and UK Sheffield) over time.
However, such trends were never universal: aboveground biomass remained unaffected by diversity in all
years in Greece (but leaf-area index increased significantly with a log-linear increase in species richness in
the third year at the Greek site) and mean biomass
declined dramatically in Portugal (with a reduced
slope). Similarly, looking at mean values the complementarity and net biodiversity effects also tended to
Results of ANOVA for decomposition of standard material (cotton and wood) in the third year.
Cotton
Source of variation
df
Site
7
Block (within site)
7
1
Species richness (log2)
Species rich. (deviation from
log-linear)
9
Functional richness (linear)
1
Functional richness (deviation)
1
7
Site 3 Species richness (log2)
Site 3 Functional richness
(linear)
7
Composition
160
24
Site 3 Composition
Residual
219
Total
443
Wood
SS
MS
F
P
SS
MS
F
P
26.8
2.2
0.6
3.8
0.3
0.6
12.2
16.8
13.5
0.002
0.000
0.000
57.6
4.7
1.2
4
4
1
0.2039
0.0075
0.0001
0.0510
0.0019
0.0001
27.1
2.5
0.2
0.004
0.045
0.685
50.8
1.9
0.0
1.5
0.3
0.2
0.3
3.8
6.8
0.000
0.010
3.1
0.6
7
1
0.0026
0.0047
0.0004
0.0047
0.6
7.8
0.742
0.006
0.6
1.2
0.0
1.5
0.0
0.2
0.0
2.1
1.000
0.089
0.1
3.2
1
4
0.0011
0.0016
0.0011
0.0004
1.8
0.9
0.180
0.492
0.3
0.4
0.1
7.1
2.5
4.1
46.6
0.0
0.0
0.1
0.0
0.1
0.4
5.6
0.994
0.999
0.000
0.3
15.2
5.4
4
114
9
140
289
0.0029
0.0688
0.0039
0.1047
0.4018
0.0007
0.0006
0.0004
0.0007
1.7
1.4
0.6
0.239
0.390
0.812
0.7
17.1
1.0
SS
%
df
SS
%
58
E. M. SPEHN ET AL.
increase with time, indicating that positive effects of
diversity on biomass are not caused only by interspecific differences in maximal growth rate. In contrast,
the selection effect was more variable between years
and sites. These results thus partly support the conceptual model of a transition from dominance of sampling effects at the beginning of biodiversity experiments to dominance of complementarity effects at later
stages, as outlined by Pacala and Tilman (2002). This
transition occurs due to changes in competitive dynamics, with dominance of exponential-growth dynamics
at very early stages and slower competitive dynamics
later on. While we saw an increase in complementarity
over time, the sampling effect was not strong at the
beginning of our experiment. Similarly, cover showed
surprisingly complex changes over time. Overall mean
values became both more positive (e.g., Ireland) and
negative (Germany, Portugal). The slope with diversity
also became steeper at some sites (e.g., Switzerland).
In some cases changes occurred only over part of the
diversity gradient: e.g., UK Silwood, where cover of
some monoculture plots declined with time, vs. UK
Sheffield, where cover of low-diversity plots increased
with time. We discuss this to point out that cover patterns were not a simple product of establishment effects
but that combinations of mortality and developmental
cycles in key species can also lead to canopy gaps, and
these effects can vary with diversity. In some cases,
mortality probably reflects expected population processes, e.g., thinning or natural decline of the initial
cohorts (effects of life history), and in others the impact
of extreme climatic events (e.g., Portugal).
Variation of diversity effects with site
At the design level our experiment is primarily focused on testing effects of diversity (the central question) and not differences due to site and species composition (which are already well known). A study focused on abiotic explanations for site differences would
need a much larger number of sites. We are also limited
in the interpretation of site effects on some ecosystem
processes as they reflect differences in both biology
and methodology (for example, soil N was not estimated at the same time within the growing season).
Nevertheless, as would be expected, site had strong
effects on all ecosystem processes. Gradients in temperature and rainfall between sites correlate with production, with reduced productivity in most northern
and most southern sites (see Hector et al. 1999: Note
20). Site differences in the impact of legumes may be
explained by levels of available phosphorous (Spehn
et al. 2002). In some cases certain sites were unusual,
e.g., Portugal experienced a strong external effect from
a dry period and particularly frosty winter. Greece
showed no or little effect of diversity on aboveground
biomass production (see also Troumbis et al. 2000) and
on most of the measured ecosystem processes, except
for cotton decomposition where it showed the strongest
Ecological Monographs
Vol. 75, No. 1
effect of species richness. The fact that diversity had
no effect on biomass at our least productive site
(Greece) and increased with aboveground production,
is consistent with Fridley (2002, 2003), who found that
effects of experimental diversity gradients on productivity significantly increased with soil fertility (but see
Kenkel et al. 2000).
Our first publication on aboveground biomass patterns from the second year of the experiment (Hector
et al. 1999) showed no significant interaction between
diversity and site, implying a single positive slope with
different intercepts for different sites (Hector et al.
1999: Fig. 1). Our multisite repeated-measures analyses of a larger range of ecosystem processes examined
over three years present a picture of broader and stronger biodiversity effects, but also revealed significant
interactions of diversity with site, time, or both on
aboveground biomass. When drawing comparisons between these results with those of Hector et al. (1999),
it needs to be considered that they include data from
the first year when the plots were establishing and patterns appear to be different mainly due to lack of time
to develop. Nevertheless, the changing patterns over
time are qualitatively consistent with Tilman et al.
(2001) who also found differences over time: an initially saturating relationship between diversity and biomass became linear in a longer-term analysis (our
changes were less strong but also over a shorter period).
More recently, Fukami and Morin (2003) have demonstrated that ecosystem functioning can be affected
by history: even under highly controlled conditions the
same set of species can achieve different productivities
depending on the order of community assembly.
Diversity effects in the context of composition
and location
In terms of the proportion of the total variation explained (i.e., the multiple R2 shown as the percentage
sums of squares explained in our tables), site and composition (plus sometimes their interaction) explained
the largest absolute portions. However, these variables
also require a lot of degrees of freedom to do it—
especially species composition with 173 degrees of
freedom. Composition is therefore a complex variable
that is hard to interpret or use for prediction (but see
Hector el al. 1999: Fig. 3, Petchey et al. 2004). In
contrast, while species richness explains less of the
sums of squares for treatments its contribution is more
parsimonious in only requiring a single degree of freedom for the regression. What is more, with diversity
on a log scale we have a simple linear relationship,
which is straightforward to use for prediction. A critical
feature of our design is that it explicitly replicated species composition (variation that is pooled into the error
term of most other biodiversity experiments because
there are no replicates for specific compositions), and
the species richness effect is tested in the context of
the composition effect by dividing the mean square for
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
species richness by the mean square for composition
in the F test. F values larger than 1 then indicate that
the average effect of species richness (per degree of
freedom) is greater than the average effect of composition. Thus, when degrees of freedom are taken into
account together with sums of squares, the average
species-richness effect (as assessed by the F value) is
90 times larger than the average species-composition
effect (for biomass). Thus, while diversity explains less
of the total variation than do site and composition it is
important to acknowledge that this relatively simple
and easy-to-obtain measure can make simple and repeatable predictions about levels of productivity and
other ecosystem processes.
Biodiversity effects and biological mechanisms
The results of our extended analyses of the selection
effect on aboveground biomass are broadly consistent
with previously published results (Loreau and Hector
2001). While the selection effect became more positive
later in the experiment it was both weaker (e.g., its
percentage sums of squares for the (log) species richness effect is one quarter that for the complementarity
effect) and more variable than the complementarity effect. Dominance through the selection effect was neither by the most productive species (the selection effect
was sometimes even negative, that is by species with
lower-than-average monoculture yields) nor was it the
main driving force behind our results. In contrast, the
complementarity effect produced one of the simplest
and most consistent results in our experiment: there
was a general positive relationship between diversity
and the complementarity effect across all years and
sites. This was evident even in the face of a wide variety
of different soil and climate conditions across the range
of sites in the experiment. Even though a variety of
different soil and climate conditions were present due
to the range of sites in the experiment, a positive complementarity effect seemed to be the rule. Analysis of
different forms of transgressive overyielding produced
qualitatively similar results to those already reported
in Hector et al. (2002a).
What biological mechanisms underlie these effects
of changes in diversity? Higher levels of vegetation
cover could explain the diversity effects in some sites,
notably in Portugal (Caldeira et al. 2001). This was due
to better establishment of some species in higher-diversity mixtures relative to lower-diversity plots and
monocultures (probably a ‘‘nurse crop’’ effect), or to
higher mortality or reduced growth in low-diversity
compositions after two or three years. However, significant effects of species richness remained after controlling for the influence of diversity via cover.
In many of our analyses additional diversity effects
remain after controlling for the effects via increased
cover, legume presence, and similar likely causes. Interactions of plants with mycorrhizas (e.g., van der Heijden et al. 1998), differentiation in rooting depth (Ber-
59
endse 1982), resource type (McKane et al. 2002), and
hydrological niche (Silvertown et al. 1999) might be
possible mechanisms for these ‘‘additional’’ biodiversity effects. The additive-partitioning analyses reveal
that both selection and complementarity effects played
a role, with the latter dominating. Niche complementarity may also occur in relation to other trophic levels
through the effects of natural enemies (see Mulder et
al. 1999, Koricheva et al. 2000, Spehn et al. 2000a,
Joshi et al. 2004).
It appears that effects of biodiversity began almost
as soon as our seeds germinated through ‘‘nurse crop’’type effects (in the first year, and at sites with many
annuals also in later years) since establishment of some
species was better in high-diversity communities. Sampling effects may then have acted through the inclusion
of fast-growing species in higher diversity communities (Huston 1997, Pacala and Tilman 2002), although
we did not see a strong influence of sampling on biomass production. The vegetation in higher diversity
communities then exploited more available space both
in two (cover) and three (canopy volume) dimensions
as well as filling space more intensively through increased canopy density aboveground and greater root
biomass and density belowground. These in turn led to
greater light interception and greater availability of N
in soil due to the additional N2 fixation in communities
with legumes. The legume effect represents both resource partitioning through use of atmospheric vs. soil
N and, later in the experiment, facilitation as fixed N
was taken up by non-legumes. While we cannot tease
all of these effects apart completely, our analyses suggest they develop at least partly in concert, rather than
as a simple, linear, causal chain. Path analysis of the
legume effect at the Swedish site revealed that initially
the simple presence or absence of legumes explained
most of the diversity effect, but later in the experiment
numbers of both legume and non-legume species became important as well as significant additional effects
of diversity per se (Mulder et al. 2002). However, path
analysis may be of limited use since we already know
the different effects of biodiversity are largely colinear
(Petraitis et al. 1996, Allison 1999, Naeem 2002). Species richness is colinear with the presence and number
of legumes in our design, therefore effects will be attributed to the single strongest effect, as we suspect
often happens when we give priority to the legumes in
analyses given that complementarity effects also occur
in experiments where legumes are not included (van
Ruijven and Berendse 2003).
The influence of diversity in communities with and
without legumes could also be examined with appropriate design constraints, namely, by having equal
numbers of mixtures with and without legumes across
the gradient of species richness. However, this would
be hard or impossible to achieve at low levels of diversity, where there may be a limited number of legume
vs. non-legume species, and at high levels of diversity,
E. M. SPEHN ET AL.
60
where species from all functional groups must be included to achieve the desired high level of diversity.
We still know relatively little about the details of competition, complementarity, and coexistence in grassland
communities. While the first generation of biodiversity
experiments has generated a lot of new information
about the relationship between diversity and ecosystem
processes, and developments in analytical methods can
tell us something about the causes, only a new generation of more mechanistic experiments will reveal
more about the underlying biology.
Conclusions
Altering biodiversity through changes in the numbers and types of plant species and functional groups
in our experimental communities significantly affected
all 11 ecosystem processes examined over the threeyear period. In general, high-diversity communities
were more productive, had stronger complementarity
effects, and exploited more resources by intercepting
more light, taking up more nitrogen, and utilizing more
2- and 3-dimensional space. These longer-term results
give greater support to niche-differentiation models as
an explanatory mechanism of the results, as complementarity was the stronger underlying biodiversity effect. Diversity effects were stronger aboveground than
belowground. In particular, clear diversity effects on
decomposition were only observed at one site. Our results therefore reinforce our previous findings and reveal stronger and more extensive effects of altered plant
diversity. They also extend the generality of diversity–
ecosystem functioning relationships to multiple sites,
years, and processes. However, because the effects of
biodiversity vary with time and site, understanding this
variation will help integrate the results of biodiversitymanipulation experiments with studies of the control
of ecosystem functioning at the larger scale.
ACKNOWLEDGMENTS
Many colleagues too numerous to list have assisted with
the project; in particular we thank Phil Heads for project
coordination and John Nelder for advice on statistical analyses. Christian Körner, Pascal Niklaus, and three anonymous
reviewers provided valuable comments on the manuscript.
The European Science Foundation (LINKECOL) funded an
exchange with the Centre of Population biology at Silwood
Park (UK), where parts of this manuscript were prepared.
Funds of the Institute of Environmental Sciences of the University of Zurich and the Botanical Institute of the University
of Basel contributed to the preparation of the manuscript. The
BIODEPTH project was funded by the European Commission
within the Framework IV Environment and Climate program
(ENV-CT95-0008) and by the Swiss Federal Office for Education and Science (Project EU-1311 to B. Schmid).
LITERATURE CITED
Allison, G. W. 1999. The implications of experimental design
for biodiversity manipulations. American Naturalist 153:
26–45.
Anonymous. 1995. Schweizerische Referenzmethoden der
Eidgenössischen landwirtschaftlichen Forschungsanstalten. Eidgenössische Forschungsanstalten für landwirtschaftlichen Pflanzenbau, Zürich, Switzerland.
Ecological Monographs
Vol. 75, No. 1
Barthram, G. T. 1986. Experimental techniques: the HFRO
swardstick. Pages 29–30 in Biennial report, 1984–1985.
Hill Farming Research Organisation Penicuik, Midlothian,
UK.
Bazzaz, F. 1996. Plants in changing environments. Cambridge University Press, Cambridge, UK.
Berendse, F. 1982. Competition between plant populations
with different rooting depths. III. Field experiments. Oecologia 53:50–55.
Bradford, M. A., et al. 2002. Impacts of soil faunal community composition on model grassland ecosystems. Science 298:615–618.
Bruno, J. F., J. J. Stachowicz, and M. D. Bertness. 2003.
Inclusion of facilitation into ecological theory. Trends in
Ecology and Evolution 18:119–125.
Brussaard, L., et al. 1997. Biodiversity and ecosystem functioning in soil. Ambio 26:563–570.
Caldeira, M. C., R. J. Ryel, J. H. Lawton, and J. S. Pereira.
2001. Mechanisms of positive biodiversity–production relationships: insights provided by d13C analysis in experimental mediterranean grassland plots. Ecology Letters 4:
439–443.
Callaway, R. M. 1995. Positive interactions among plants.
Botanical Review 61:306–348.
Compton, J. E., and R. D. Boone. 2000. Long-term impacts
of agriculture on soil carbon and nitrogen in New England
forests. Ecology 81:2314–2330.
Cropp, R., and A. Gabric. 2002. Ecosystem adaptation: do
ecosystems maximize resilience? Ecology 83:2019–2036.
Diaz, S., A. J. Symstad, F. S. Chapin, III, D. A. Wardle, and
L. F. Huenneke. 2003. Functional diversity revealed by
removal experiments. Trends in Ecology and Evolution 18:
140–146.
Diemer, M., J. Joshi, C. Körner, B. Schmid, and E. Spehn.
1997. An experimental protocol to assess the effects of
plant diversity on ecosystem functioning utilized in a European research network. Bulletin of the Geobotanical Institute ETH 63:95–107.
Diemer, M., and B. Schmid. 2001. Effects of biodiversity
loss and disturbance on the survival and performance of
two Ranunculus species with differing clonal architectures.
Ecography 24:59–67.
Elashoff, J. D. 1986. Analysis of repeated measures designs.
BMDP technical report, number 83. BMDP Statistical Software, Los Angeles, California USA.
Emmerson, M. C., M. Solan, C. Emes, D. M. Peterson, and
D. Raffaelli. 2001. Consistent patterns and the idiosyncratic effects of biodiversity in marine ecosystems. Nature
411:73–77.
Ewel, J., M. Mazzarino, and C. Berish. 1991. Tropical soil
fertility changes under monocultures and successional communities of different structure. Ecological Applications 1:
289–302.
Fargione, J., and D. Tilman. 2002. Competition and coexistence in terrestrial plants. Pages 165– 206 in U. Sommer
and B. Worm, editors. Competition and coexistence.
Springer-Verlag, Berlin, Germany.
Fridley, J. D. 2002. Resource availability dominates and alters the relationship between species diversity and ecosystem productivity in experimental plant communities. Oecologia 132:271–277.
Fridley, J. D. 2003. Diversity effects on production in different light and fertility environments: an experiment with
communities of annual plants. Journal of Ecology 91:396–
406.
Fukami, T., and P. J. Morin. 2003. Productivity–biodiversity
relationships depend on the history of community assembly. Nature 424:423–426.
Gastine, A., M. Scherer-Lorenzen, and P. W. Leadley. 2003.
No consistent effect of plant diversity on root biomass, soil
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
biota and soil abiotic conditions in temperate grassland
communities. Applied Soil Ecology 24:101–111.
Givnish, T. J. 1994. Does diversity beget stability? Nature
371:113–114.
Grace, J. B. 1999. The factors controlling species density in
herbaceous plant communities: an assessment. Perspectives
in Plant Ecology, Evolution and Systematics 2:1–28.
Green, B. F., and J. W. Tukey. 1960. Complex analysis of
variance: general problems. Psychometrika 25:127–152.
Grime, J. P. 1998. Benefits of plant diversity to ecosystems:
immediate, filter and founder effects. Journal of Ecology
86:902–910.
Harper, J. L. 1977. Population biology of plants. Academic
Press, London, UK.
He, J.-S., F. A. Bazzaz, and B. Schmid. 2002. Interactive
effects of diversity, nutrients and elevated CO2 on experimental plant communities. Oikos 97:337–348.
Hector, A. 1998. The effect of diversity on productivity: detecting the role of species complementarity. Oikos 82:597–
599.
Hector, A., E. Bazeley-White, M. Loreau, S. Otway, and B.
Schmid. 2002a. Overyielding in plant communities: testing
the sampling effect hypothesis with replicated biodiversity
experiments. Ecology Letters 5:502–511.
Hector, A., A. J. Beale, A. Minns, S. J. Otway, and J. H.
Lawton. 2000a. Consequences for the reduction of plant
diversity for litter decomposition: effects through litter
quality and microenvironment. Oikos 90:357–371.
Hector, A., K. Dobson, A. Minns, E. Bazeley-White, and J.
H. Lawton. 2001a. Community diversity and invasion resistance: an experimental test in a grassland ecosystem and
a review of comparable studies. Ecological Research 16:
819–831.
Hector, A., J. Joshi, S. P. Lawler, E. Spehn, and A. Wilby.
2001b. Conservation implications of the link between biodiversity and ecosystem functioning. Oecologia 129:624–
628.
Hector, A., M. Loreau, and B. Schmid. 2002b. Biodiversity
manipulation experiments: studies replicated at multiple
sites. Pages 36–46 in M. Loreau, S. Naeem, and P. Inchausti, editors. Biodiversity and ecosystem functioning:
synthesis and perspectives. Oxford University Press, Oxford, UK.
Hector, A., et al. 1999. Plant diversity and productivity experiments in European grasslands. Science 286:1123–1127.
Hector, A., et al. 2000b. No consistent effect of plant diversity on productivity? Response. Science 289:1255a.
Hirose, T., and M. J. A. Werger. 1994. Photosythetic capacity
and nitrogen partitioning among species in the canopy of
a herbaceous plant community. Oecologia 100:203–212.
Hooper, D. U., et al. 2000. Interactions between aboveground
and belowground biodiversity in terrestrial ecosystems:
patterns, mechanisms and feedbacks. BioScience 50:1049–
1061.
Hooper, D. U., and P. M. Vitousek. 1997. The effects of plant
composition and diversity on ecosystem processes. Science
277:1302–1305.
Hooper, D. U., and P. M. Vitousek. 1998. Effects of plant
composition and diversity on nutrient cycling. Ecological
Monographs 68:121–149.
Huston, M. A. 1997. Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. Oecologia 110:449–460.
Huston, M. A., L. W. Aarssen, M. P. Austin, B. S. Cade, J.
D. Fridley, E. Garnier, J. P. Grime, J. Hodgson, W. K. Laurenroth, K. Thompson, J. H. Vandemeer, and D. A. Wardle.
2000. No consistent effect of plant diversity on productivity? Science 289:1255.
Huston, M. A., and A. C. McBride. 2002. Evaluating the
relative strengths of biotic versus abiotic controls on eco-
61
system processes. Pages 36–46 in M. Loreau, S. Naeem,
and P. Inchausti, editors. Biodiversity and ecosystem functioning: synthesis and perspectives. Oxford University
Press, Oxford, UK.
Joshi, J., D. Matthies, and B. Schmid. 2000. Root hemiparasites and plant diversity in experimental grassland communities. Journal of Ecology 88:634–644.
Joshi, J., S. Otway, J. Koricheva, A. B. Pfisterer, J. Alphei,
B. A. Roy, M. Scherer-Lorenzen, B. Schmid, E. Spehn, and
A. Hector. 2004. Bottom-up effects and feed-backs in simple and diverse experimental grassland communities. Pages
115–134 in W. Weisser and E. Siemann, editors. Insects
and ecosystem function. Springer-Verlag, Berlin, Germany.
Joshi, J., et al. 2001. Local adaptation enhances performance
of common plant species. Ecology Letters 4:536–544.
Jumpponen, A., P. Högberg, K. Huss-Danell, and C. P. H.
Mulder. 2002. Interspecific and spatial differences in nitrogen uptake in monocultures and two-species mixtures in
north European grasslands. Functional Ecology 16:454–
461.
Kenkel, N. C., D. A. Peltzer, D. Baluta, and D. Pirie. 2000.
Increasing plant diversity does not influence productivity:
empirical evidence and potential mechanisms. Community
Ecology 1:165–170.
Kinzig, A., D. Tilman, and S. Pacala, editors. 2002. The
functional consequences of biodiversity: empirical progress
and theoretical extensions. Princeton University Press,
Princeton, New Jersey, USA.
Koricheva, J., C. P. H. Mulder, B. Schmid, J. Joshi, and K.
Huss-Danell. 2000. Numerical responses of different trophic groups of invertebrates to manipulations of plant diversity in grasslands. Oecologia 126:310–320.
Lawton, J. H. 2000. Community ecology in a changing world.
Ecology Institute, Oldendorf/Luhe, Germany.
Levin, S. A. 1997. Biodiversity: interfacing populations and
ecosystems. Pages 277–288 in T. Abe, S. A. Levin, and M.
Higashi, editors. Biodiversity: an ecological perspective.
Springer-Verlag, New York, New York, USA.
Loreau, M. 1998. Biodiversity and ecosystem functioning: a
mechanistic model. Proceedings of the National Academy
of Sciences (USA) 95:5632–5636.
Loreau, M. 2000. Biodiversity and ecosystem functioning:
recent theoretical advances. Oikos 91:3–17.
Loreau, M., and A. Hector. 2001. Partitioning selection and
complementarity in biodiversity experiments. Nature 412:
72–76 [erratum: 413:548].
Loreau, M., S. Naeem, and P. Inchausti, editors. 2002. Biodiversity and ecosystem functioning: synthesis and perspectives. Oxford University Press, Oxford, UK.
Loreau, M., S. Naeem, P. Inchausti, J. Bengtsson, J. P. Grime,
A. Hector, D. U. Hooper, M. A. Huston, D. Raffaelli, B.
Schmid, D. Tilman, and D. A. Wardle. 2001. Biodiversity
and ecosystem functioning: current knowledge and future
challenges. Science 294:804–809.
McKane, R. B., L. C. Johnson, G. R. Shaver, K. J. Nadelhoffer, E. B. Rastetter, B. Fry, A. E. Giblin, K. Kielland,
B. L. Kwiatkowski, J. A. Laundre, and G. Murray. 2002.
Resource-based niches provide a basis for plant species
diversity and dominance in arctic tundra. Nature 415:68–
71.
McNaughton, S. J. 1993. Biodiversity and function of grazing
ecosystems. Pages 362–383 in E.-D. Shulze and H. A. Mooney, editors. Biodiversity and ecosystem function. Springer-Verlag, Berlin, Germany.
Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross,
H. L. Reynolds, R. B. Waide, M. R. Willig, M. R. Dodson,
and L. Gough. 2001. What is the observed relationship
between species richness and productivity? Ecology 82:
2381–2396.
62
E. M. SPEHN ET AL.
Moran, M. D. 2003. Arguments for rejecting the sequential
Bonferonni in ecological studies. Oikos 100:403–405.
Mulder, C. P. H., A. Jumpponen, P. Högberg, and K. HussDanell. 2002. How plant diversity and legumes affect nitrogen dynamics in experimental grassland communities.
Oecologia 133:412–421.
Mulder, C. P. H., J. Koricheva, K. Huss-Danell, P. Högberg,
and J. Joshi. 1999. Insects affect relationships between
plant species richness and ecosystem processes. Ecology
Letters 2:237–246.
Mulder, C. P. H., D. D. Uliassi, and D. F. Doak. 2001. Physical
stress and diversity–productivity relationship: the role of
positive species interactions. Proceeding of the National
Academy of Sciences (USA) 98:6704–6708.
Naeem, S. 2002. Disentangling the impacts of diversity on
ecosystem functioning in combinatorial experiments. Ecology 83:2925–2935.
Naeem, S., L. J. Thompson, S. P. Lawler, J. H. Lawton, and
R. M. Woodfin. 1994. Declining biodiversity can alter the
performance of ecosystems. Nature 368:734–737.
Naeem, S., L. J. Thompson, S. P. Lawler, J. H. Lawton, and
R. M. Woodfin. 1995. Empirical evidence that declining
species diversity may alter the performance of terrestrial
ecosystems. Philosophical Transactions of the Royal Society of London B 347:249–262.
Neter, J., and W. Wasserman. 1974. Applied linear statistical
models. Richard D. Irwin, Homewood, Illinois, USA.
Niklaus, P. A., P. W. Leadley, B. Schmid, and C. Korner. 2001.
A long-term study on biodiversity 3 elevated CO2 interactions in grassland. Ecological Monographs 71:341–356.
Nilsson, M.-C., D. A. Wardle, and A. Dahlberg. 1999. Effects
of plant litter, species composition, and diversity on the
boreal forest plant–soil system. Oikos 86:16–26.
Pacala, S., and D. Tilman. 2002. The transition from sampling
to complementarity. Pages 151–166 in A. Kinzig, D. Tilman, and S. Pacala, editors. Functional consequences of
biodiversity: experimental progress and theoretical extensions. Princeton University Press, Princeton, New Jersey
USA.
Payne, R. W., P. W. Lane, P. G. N. Digby, S. A. Harding, P.
K. Leech, G. W. Morgan, A. D. Todd, R. Thompson, G.
Tunicliffe Wilson, S. J. Welham, and R. P. White. 1993.
GENSTAT 5 reference manual. Clarendon Press, Oxford,
UK.
Petchey, O. L., A. Hector, and K. J. Gaston. 2004. How do
different measures of functional diversity perform? Ecology 85:847–857.
Petraitis, P. S., A. E. Dunham, and P. H. Niewiarowski. 1996.
Inferring multiple causality: the limitations of path analysis. Functional Ecology 10:421–431.
Pfisterer, A. B., J. Joshi, B. Schmid, and M. Fischer. 2004.
Rapid decay of diversity–productivity relationships after
invasion of experimental plant communities. Basic and Applied Ecology, 5:5–14.
Pfisterer, A. B., and B. Schmid. 2002. Diversity-dependent
production can decrease the stability of ecosystem functioning. Nature 416:84–86.
Reich, P. B., et al. 2001. Plant diversity enhances ecosystem
responses to elevated CO2 and nitrogen deposition. Nature
410:809–812.
Rice, W. R. 1989. Analysing tables of statistical tests. Evolution 43:223–225.
Rosenthal, R., and R. L. Rosnow. 1985. Contrast analysis:
focused comparisons in the analysis of variance. Cambridge University Press, Cambridge, UK.
Scherer-Lorenzen, M. 1999. Effects of plant diversity on ecosystem processes in experimental grassland communities.
Bayreuther Forum Ökologie 75:1–195.
Scherer-Lorenzen, M., C. Palmborg, A. Prinz, and E.-D.
Schulze. 2003. The role of plant diversity and composition
for nitrate leaching in grasslands. Ecology 84:1539–1552.
Ecological Monographs
Vol. 75, No. 1
Schmid, B., A. Hector, M. A. Huston, P. Inchausti, I. Nijs, P.
W. Leadley, and D. Tilman. 2002a. The design and analysis
of biodiversity experiments. Pages 61–75 in M. Loreau, S.
Naeem, and P. Inchausti, editors. Biodiversity and ecosystem functioning. Oxford University Press, Oxford, UK.
Schmid, B., J. Joshi, and F. Schläpfer. 2002b. Empirical evidence for biodiversity–ecosystem functioning relationships. Pages 120–150 in A. Kinzig, D. Tilman, and S. P.
Pacala, editors. Functional consequences of biodiversity:
experimental progress and theoretical extensions. Princeton
University Press, Princeton, New Jersey, USA.
Schulze, E.-D., R. Leuning, and F. M. Kelliher. 1995. Environmental regulation of surface conductance for evaporation from vegetation. Vegetatio 121:79–87.
Schulze, E.-D., and H. A. Mooney, editors. 1993. Biodiversity and ecosystem function. Springer-Verlag, Berlin, Germany.
Schwinning, S., and A. J. Parsons. 1996. A spatially explicit
population model of stoloniferous N-fixing legumes in
mixed pasture with grass. Journal of Ecology 84:815–826.
Silvertown, J., M. E. Dodd, D. J. G. Gowings, and J. O.
Mountford. 1999. Hydrologically defined niches reveal a
basis for species richness in plant communities. Nature 400:
61–63.
Spehn, E. M., J. Joshi, B. Schmid, J. Alphei, and C. Körner.
2000a. Plant diversity effects on soil heterotrophic activity
in experimental grassland ecosystems. Plant and Soil 224:
217–230.
Spehn, E. M., J. Joshi, B. Schmid, M. Diemer, and C. Körner.
2000b. Aboveground resource use increases with plant species richness in experimental grassland ecosystems. Functional Ecology 14:326–337.
Spehn, E. M., et al. 2002. The role of legumes as a component of biodiversity in a cross-European study of grassland biomass nitrogen. Oikos 98:205–218.
Stephan, A., A. H. Meyer, and B. Schmid. 2000. Plant diversity affects culturable soil bacteria in experimental
grassland communities. Journal of Ecology 88:988–998.
Symstad, A. J., D. Tilman, J. Willson, and J. M. H. Knops.
1998. Species loss and ecosystem functioning: effects of
species identity and community composition. Oikos 81:
389–397.
Tilman, D. 1988. Plant strategies and the dynamics and structure of plant communities. Princeton University Press,
Princeton, New Jersey, USA.
Tilman, D. 1994. Competition and biodiversity in spatially
structured habitats. Ecology 75:2–16.
Tilman, D., J. Knops, D. Wedin, and P. Reich. 2002. Plant
diversity and composition: effects on producivity and nutrient dynamics of experimental grasslands. Pages 21–35
in M. Loreau, S. Naeem, and P. Inchausti, editors. Biodiversity and ecosystem functioning: synthesis and perspectives. Oxford University Press, Oxford, UK.
Tilman, D., J. Knops, D. Wedin, P. Reich, M. Ritchie, and E.
Siemann. 1997a. The influence of functional diversity and
composition on ecosystem processes. Science 277:1300–
1302.
Tilman, D., C. L. Lehman, and K. T. Thomson. 1997b. Plant
diversity and ecosystem productivity: theoretical considerations. Proceedings of the National Academy of Science
(USA) 94:1857–1861.
Tilman, D., P. B. Reich, J. M. H. Knops, D. Wedin, T. Mielke,
and C. Lehman. 2001. Diversity and productivity in a longterm grassland experiment. Science 294:843–845.
Tilman, D., D. Wedin, and J. Knops. 1996. Productivity and
sustainability influenced by biodiversity in grassland ecosystems. Nature 379:718–720.
Troumbis, A. Y., P. G. Dimitrakopoulos, A.-S. D. Siamantziouras, and D. Memtsas. 2000. Hidden diversity and pro-
February 2005
ECOSYSTEM EFFECTS OF PLANT DIVERSITY
ductivity patterns in mixed Mediterranean grasslands. Oikos 90:549–559.
Troumbis, A. Y., A. Galanidis, and G. D. Kokkoris. 2002.
Components of short-term invasibility in experimental
Mediterranean grasslands. Oikos 98:239–250.
Van der Heijden, M. G. A., J. N. Klironomos, M. Ursic, P.
Moutoglis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and
I. R. Sanders. 1998. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability, and productivity. Nature 396:69–72.
Van der Putten, W. H., et al. 2000. Plant species diversity
as a driver of early succession in abandoned fields: a multisite approach. Oecologia 124:91–99.
van Ruijven, J., and F. Berendse. 2003. Positive effects of
plant species diversity on productivity in the absence of
legumes. Ecology Letters 6:170–175.
Wardle, D. A. 2002. Communities and ecosystems; linking
the aboveground and belowground components. Princeton
University Press, Princeton, New Jersey, USA.
63
Wardle, D. A., K. I. Bonner, G. M. Barker, G. W. Yeates, K.
S. Nicholson, R. D. Bardgett, R. N. Watson, and A. Ghani.
1999. Plant removals in perennial grassland: vegetation
dynamics, decomposers, soil biodiversity, and ecosystem
properties. Ecological Mongraphs 69:535–568.
Wardle, D. A., K. I. Bonner, and K. S. Nicholson. 1997.
Biodiversity and plant litter: experimental evidence which
does not support the view that enhanced species richness
improves ecosystem function. Oikos 79:247–258.
Werger, M. J. A., and T. Hirose. 1988. Effects of light climate
and nitrogen partitioning on the canopy structure of stands
of a dicotyledonous, herbaceous vegetation. Pages 171–181
in M. J. A. Werger, P. J. M. Van der Aart, H. J. During,
and J. T. A. Verhoeven, editors. Plant form and vegetation
structure. SPB Academic Publishing, The Hague, The
Netherlands.
Zak, D. R., W. E. Holmes, D. C. White, A. D. Peacock, and
D. Tilman. 2003. Plant diversity, soil microbial communities, and ecosystem function: are there any links? Ecology
84:2042–2050.
SUPPLEMENT
A supplement containing the data from the BIODEPTH project (15 ecosystem-process variables measured at eight different
European grassland field sites over three years) together with metadata and a table with site information is available in ESA’s
Electronic Data Archive: Ecological Archives M075-001-S1.